From Selenium 4.8.0, support for non-w3c browser control has ended.
We only use W3C browser control these days, and this was missed as part
of the move to W3C. All browser options must be vendor-prefixed.
We should be proactive in ensuring that the environment is clean when
running a task. We already ensure that we have a clean renderer and
other parts of the output chain, but we were not setting a clean user.
This change adds a call to setup the cron user before each task is
actually executed.
The main goal of this issue is to avoid scanners (Dependabot
and friends), reporting about security issues with the current
xmldom 0.6.0 package.
Note that this doesn't affect prod at all, because it's a dev
dependency, hardly exploitable. So it's not a security fix, just
a security_benefit, if something.
So here, we are updating from xmldom 0.6.0 to @xmldom/xmldom 0.8.7
(note that the package was renamed in 0.7.0, so it's the very same)
Also, when proceeding with the changes, it was detected that we
are incorrectly declaring @babel/eslint-parser as a normal dependency
instead of a development one, so we are also fixing that little detail.
The commands executed to get the changes above applied have been:
- nvm use
- npm install @xmldom/xmldom@^0.8.7 --save-dev
- npm uninstall xmldom
- npm install @babel/eslint-parser@^7.17.0 --save-dev
(we haven't run a complete re-install because we only want to modify
the minimum possible at this stage).
MDL-76499 revealed a few problems with resource generators:
1. We were not covering with unit tests the upload of files from disk
(and here it's where the problem was).
2. There was a little of confusion between disk paths (only needed
to upload files) and file_area paths (the generator only creates
or uploads files to the root directory of the file area.
3. It was possible to request the upload of a file to the generator
without that file effectively existing.
This commit fixes those points and covers 99% of the generator code.
extrainfo is an extra information for the DB driver, e.g. SQL Server,
has additional configuration according to its environment,
which the administrator can specify to alter and override any connection options.
Co-authored-by: LukeCarrier <luke@carrier.im>
This is a backport of MDL-64153.
When a test runs just after midnight, and the user time zone is not the
same as the server timezone, and the course is created using a generator
(which runs in server time zone) but the UI presented in the user
timezone, the course start time is still in the future.
We need to create the course a day earlier to ensure that the "This
week" indicator is in the correct day.
This commit switches to a new custom version of Shifter which sees it
put on a massive diet.
Many of the standard Shifter features are totally unused by Moodle, but
they add dependency spaghetti. Some of those dependencies will start to
cause us issues as they have not been updated for a very long time, are
no longer maintained, and are no longer compatible with anything
resembling a recent version of NodeJS.
This resolves an issue identified in 77083 which identified that GitHub
hashes shouldn't be treated as repeatable.
Whilst GitHub have reverted this change, it's still worth making this
change.
Allow the repository call to the web service method to complete before
re-directing, otherwise the call may not complete and/or races against
the page reload causing random Behat errors.
Right now, any E_DEPRECATED, E_USER_DEPRECATED error type leads
to exception in behat runs, because the behat error handled is
not aware of them. See MDL-38041 for the initial implementation.
We don't want that to happen and they must behave like other "low"
error types (notices, warnings...), that will be reported in logs
or output, but won't trigger any exception, so the test can continue.
This commit just adds those 2 "new" error types to the handler,
providing the very same behaviour for them than the rest.
Replaced 4 occurrences of get_debug_type() by gettype() to keep PHP 7.4 compatibility. Note this
has not been applied to 4.2dev and up because, there, it's safe to use get_debug_type().
Replace steps that manually add resource instances via the UI and use
Behat generators. This improves the speed of the Behat test runs.
Co-authored-by: David Woloszyn <david.woloszyn@moodle.com>
Replace steps that manually add Book instances via the UI and use
Behat generators. This improves the speed of the Behat test runs.
Additionally, performed Behat cleanup and optimisation to affected
files.
This is because on classic theme the navigation block pushes the content
to the right and once behat changes the grades the 'Save changes' button
is no longer visible.
It happens often that some timezone is renamed, split, added to the
IANAs list of timezones (that, for PHP is kept updated by the
php-timezonedb PECL extension).
When the information coming from the extension changes, all the
PHPUnit jobs in the world start failing (that's when CIs update
their PHP images, when devs update their packages, ...).
So, what we are doing here is to move that test that check for
every single string existing to be run only when PHPUNIT_LONGTEST
is enabled. That way only places running all tests will run that
one. And every other run just will skip it.
This corresponds to point 1 & 2 of the issue, lower impact in general
and, still be able to enable the tests to run.
By using that assertion the arrays will be sorted before comparison
preventing any Oracle sorting related failures.
Credits for the solution go to Tim Hunt!
This will find all data generators that can be used in behat via the
'the following "something" exist:' step, and display them in a select
list on the step definitions page.
When a generator is selected, it will fetch the required fields for that
generator and display them on the page.
This change reduces the priority of the shipped phpcs configuration to
allow for local overrides.
This is beneficial in several situations:
- when users wish to apply stricter or additional standards
- when the integration team wishes to trial new rules
Under some linux versions, and depending of the configured
locale categories, it's possible to get a current locally
which length > 255 when calling to setlocale(LC_ALL, 0).
Later, if that long locale is tried to be restored, there
is a "setlocale(): Specified locale name is too long" warning
error.
When that happens we need to split the long locale into
individual chunks and set all the (six) locale categories
supported one by one.
Covered with tests, note that, in practice, this only
happens with linux because it supports 12 locale categories
@ OS level. Both BSD (6) and Windows (5) hardly can reach the limit.
No matter of that, the tests have been designed to ensure that
they pass on all OSs, just the new code only will be executed
on linux.
This commit adds an additional environment check for the unsupported
PHP 8.1 version to Moodle 3.9.x, 3.11.x and 4.0.x.
It also updates the lang string for unsupported PHP version.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
This fix is really only required for Moodle 4.1 and up, that will
be the one supporting php81, but we always apply composer changes
to all branches (supported and security), so we keep them running
the same stuff.
Note this comes with a bunch of other small updates, all them
pretty minor and should not be problematic. Better we get them
than keeping them apart.
Updates the Aiken Format class to process answers correctly, enabling
support for special HTML characteres such as <, >, and &.
Co-authored-by: Leticia Luz <leticia.adrielli.luz@gmail.com>
- The current item should be focusable otherwise the focus will be lost
as soon as a user clicks on an item in the initials bar
- Modified behat tests to not mistakenly click on the 'All' link
- The currently selected item is specified by aria-current
- Useed aria-current=true rather than =page because the links of each
item in the initials bar is missing the pagination information
If member sync runs before the user launches the tool, a partial record
is created, without consumer secret. Subsequent launches of the tool by
that member don't resolve this and this results in grade sync failing
for any affected users. This patch:
- data fixes the existing affected rows
- fixes the launch code, ensuring secret is recorded on launch,
irrespective of whether the user info record has been created already
or not.
For Moodle 4.1 and up, the php-xmlrpc is not needed anymore:
- All the MNet stuff has been moved to use php library (MDL-76055).
- The webservice/xmlrpc has been moved from core to contrib (MDL-76052).
So we just remove the check here. Starting with 4.1, it's not
needed for any core functionality.
No lang string changes in this branch, only in master (4.1dev).
Because @coversDefaultClass doesn't indicate any coverage (it's
just an alias to avoid having to write the class name in @covers),
this commit fixes all the files that were using that annotation
and missing any @covers.
Basically, replacing one by the other and done.
* When the language pack being deleted is
- the same as the site default language, we must set the site default
language to 'en'.
- fix the user's current language to the default site language.
These steps have accepted a NodeElement instance as an argument for some
time, but were trying to cast it to string when formulating exception
messages, making it harder to debug and, in the case of the 'should see'
step, not work at all.
This patch introduces a new function to produce a consistent naming for
them.
This commit adds new steps related to action menus to support:
* Choosing an item in a named action menu
* Choosing an item in a named action menu within a container
* Confirming that an action menu item does or does not exist in a named
action menu
* Confirming that an action menu item does or does not exist in a named
action menu within a container
The existing action menu steps were insufficient as they assume that
there is only one action menu within he container, which is not
necessarily the case.
The existing action menu steps are not non-JS friendly and will error if
JS is disabled, without providing any fallback when one is easily
available.
Unfortunately these steps cannot be used to replace the existing steps
without manual intervention.
Fixes the enrol_lti_users.consumersecret field for LTI 2.0 users.
This field erroneously contained the tool secret and not the consumer
secret needed for service requests when used with LTI 2.0 consumers,
which resulted in complete grade sync failure for LTI 2.0 consumers.
This patch:
- adds an upgrade step to address existing incorrect secrets for LTI
2.0 launched users. It sets these to the correct consumer secret.
- fixes the way the secret is first set during a launch, ensuring
this->consumer->secret is used, which properly captures either the
tool secret (for 1.1 launches) or the consumer secret (for 2.0
launches).
Adjusts the testing scenarios so that we don't rely on endpoint
discovery being run during an issuer edit, which is no longer the case.
For scenarios that need a userinfo endpoint, set this manually.
In cases where we have an id, load the persistent and then set new data
before saving. This ensures we don't delete fields which are missing in
the form data.
* The setting "Display mode" modifies the choices alignment of radio choice button
from horizontally to vertically. It should also display the list of choices in
the graph in the same directions.
This commit addresses a race condition when showing the message drawer
and opening the settings page from the Message preferences page.
In that situation, the message/edit.php page calls the
message_drawer_helper.showSettings function, which publishes a request
to show the drawer.
In some conditions (notably faster machines, or after MDL-66107) this
event is fired before the message_drawer.init has listened to it.
The solution here is to delay firing of the event until the drawer
announces itself as ready.
This ensures that the event is subscribed to before it is fired.
Polling conversions will run the document conversion immediately if the
conversion has not already been completed.
Releasing the session lock while doing this ensures large conversions
do not unnecessarily hold the users session.
The lock ensures that multiple conversions do not happen for the same
assignment submission.
Otherwise, subsequent conversions will fail when trying to save the
converted file.
The mnet_environment->keypair array contains the following
elements (and more, just focussing on these):
- keypair_PEM : textual representation of the private key.
- certificate : textual representation of the public key.
- privatekey : OpenSSLAsymmetricKey representation of the private key,
generated from keypair_PEM. See get_private_key().
- publickey : OpenSSLAsymmetricKey representation if the public key,
generated from certificate. See get_public_key().
The last 2 elements in the array are only used as "caching", to avoid
having to call to openssl_pkey_get_private() and
openssl_pkey_get_public() to convert from the textual representation
to the OpenSSLAsymmetricKey representation that is the one required
by a number of openssl functions.
Problems arrive when, as part of the MNet protocol, the mnet_environment
is serialised, because, since PHP 8.0 those OpenSSLAsymmetricKey objects
aren't serialisable any more.
So, as far as they are only used for internal caching it's perfectly ok
to remove the caching bits and use the openssl_pkey_get_xxx() methods
to calculate them under demand.
The alternative to this would be to implement into the mnet_environment
some custom serialisation, skipping those OpenSSLAsymmetricKey
instances, using __sleep(), the Serializabla interface or __serialize(),
but that seems unnecessary because, as explained above, the uses are
really limited and easily replaceable.
That's what this patch does.
This is basically the manual background that was removed in the
previous commit, with 4 attempts here. Note it includes the 2
waits that are REQUIRED to make it pass in too-quick environments.
If they are removed, they begin to fail because it seems that
the attempts need a minimum of separation between them.
This moves away from manual attempts and uses the activity
attempts generator instead. Quicker and not exposed to the
problems with clicks, refreshes and missing attempts.
Note that there is a small change in the attempt generator
so, now, when a new attempt is created, if the activity has
tracking enabled, the final result / grades are recalculated.
That's the real behaviour when the activity settings are
changes in the edition form and it's natural to get them
recalculated without having to visit the form. Specifically,
this allows the 1st scenario to pass, because in that scenario
no changes to the form are being performed.
Also, little detail, amend a dupe scenario name.
This commit adds a new Grunt task which generates a jsconfig.json file.
This file is described at
https://code.visualstudio.com/docs/languages/jsconfig and can be used
to configure vscode and other similar editors to understand our module
loading.
This task is an optional task for Grunt and can be run with:
npx grunt jsconfig
The file is git-ignored and will not affect normal usage.
Some editors do not immediately store their content to the textarea they
represent and only do so when the form is submitted or they are
requested to.
This change adds to the existing API to allow an (optional)
`store_current_value` function to be defined for the editor type so that
it can use the JavaScript API to persist any current content to the
editor.
This commit adds a standardise_html function and updates the matches
function to compare normalised content.
This allows for a wider variety of valid editor output to be handled
using the standard value matching steps in Behat, thus supporting
editors other than Atto better.
Prior to this commit, the behat_form_editor integration was hard-coded
to use Atto.
This change allows other editors to be used to set editor values, and
breaks that hard bond. Following this change, any editor is able to
define a behat_editor_[name].php file defining a function which meets
the following signature:
set_editor_value(string $editorid, string $value): void
Each editor is responsible for checking whether its own API is available
and calling any relevant functions to force the editor to set the
content.
Please note: Behat is unable to determine the current editor in use on
the page automatically.
These steps are currently using Atto-specific UI features and must
therefore make use of the @editor_atto tag to ensure that they continue
to run using Atto.
A new issue will be raised to migrate each of these to be
editor-agnostic.
When an editor is used in a form as part of a group, it is missing its
label element.
This happens because there is no inline template for the editor, and
therefore it reverts to using the old `toHtml` function. The legacy
function does not cater to the editor being in a group, and therefore
does not add the hidden label.
In addition, the hiddenLabel attribute was missing from the editor
element. This is a semi-standard element but must be added to each
supporting element.
Both the missing inline template for the editor, and the hidden label
attribute must be present:
- If the hidden label attribute is not present, then the standard inline
template will add a visible label within the group.
- If the inline editor template is missing, then the label is not shown
at all.
When the template editor is configured to use the HTML Editor, it should
indicate this to Behat by setting the `data-fieldtype` to "editor".
This indicates to the behat field manager that the editor is in use,
which ensures that the correct value setting can take place.
This is important in the context of MDL-75887.
This change, which should be easy to mimic for other editors.
This will ensure that the correct editor is used for tests relating to
that editor, or its subplugins.
The OR conditions in the WHERE clause prevented the query from
effectively filtering the messages related to the user quickly, this
change helps gets around this by allowing the database to limit
the rows in the messages table it needs to scan significantly.
The following change adds support to show only active course users
in the forum grader and generally replicates the behaviour from the
assignment grader which relies on certain config settings, user
permissions and capabilities.
Modifies the 'core_course_get_enrolled_users_by_cmid' webservice and
the generic js methods that call this webservice and enables defining
whether the ws should only return the active users or all enrolled
users in the course.
This issue just goes over all the currently incorrect
namespaces in test cases and:
1. Change the namespace to the correct one.
2. Move/rename it to correct location if needed (level 2 and down).
3. Remove not needed MOODLE_INTERNAL check when possible.
4. Remove file phpdoc when the file only has one class.
5. Make small adjustments in case the change of namespace requires it.
Have tried to also bump Travis to 22.04 (jammy) but it
was not working ok, with problems trying to find both
PHP and PostgreSQL versions.
Surely that's because 22.04 support is really new (Sep 9, 2022):
https://blog.travis-ci.com/22-9-9-ubuntujammy
So we'll have to wait a little more before being able to, also,
bump Travis to use 22.04
In stables, just minors updates to components.
Generated using the instructions @ https://docs.moodle.org/dev/Composer
and using PHP 7.3 (always the min version supported) to generate the
lock file.
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Recently, PHPUnit (all versions) did some changes about how the
comparators worked and this has caused some float comparisons
to stop working.
We need to move them to assertEqualsWithDelta(), allowing a small
tolerance to workaround the floats comparison problem.
0.00001 has been decided. And applied to all the similar assertions
within the unittest function, so if more cases are added to them
better be copied with the agreed tolerance.
The cache file is only ever written to in the `write_file` function,
where it does so by writing to a temp file and performing an atomic
rename of that file. When writing, the target file is never locked.
The cache file is only ever read in the `get` function, and there is no
need for an exclusive lock in that situation.
There is therefore no need to obtain any lock, shared or exclusive, to
read the cache file. Doing so only affects performance of the file sytem
as file locks must be needlessly obtained and written to disk for a read
operation which does not benefit from them.
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- In lib/tests/blocklib_test.php 2 helper classes have been
moved to tests/fixtures, because they needed to be namespace-free.
Calling \core_user\fields::get_sql() incerements a static uniqueid
variable. This should be called only once and not make a separate call
for the sort mapping data as it doing so will result to an incorrect
table alias for the sort mapping data.
With IANA TZDB (timezone database) release 2022b, Europe/Kiev
has been renamed to Europe/Kyiv. So we need to add this string
to our list to keep tests passing.
Note that the old Europe/Kiev continues working (is an alias) and,
also, there are places in core where we are using it. Just we cannot
remove/replace those uses in core until we ensure that everybody is
using PHP 8.0.23 (TZDB 2022b) or more recent versions. Hence,
not changing those uses now.
Mustache version 2.14.1 is already included, but the VERSION const
incorrectly listed 2.14.0 in the tagged Mustache release. A later
upstream commit corrected this in the library, which this now matches.
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- The following task tests have been moved within the level2 directory:
- \core\adhoc_task_test => \core\task\adhoc_task_test
- \core\scheduled_task_test => \core\task\scheduled_task_test
- \core\calendar_cron_task_test => \core\task\calendar_cron_task_test
- \core\h5p_get_content_types_task_test => \core\task\h5p_get_content_types_task_test
- \core\task_database_logger_test => \core\task\database_logger_test
- \core\task_logging_test => \core\task\logging_test
- The following event tests have been moved within level2 directory:
- \core\event_context_locked_test => \core\event\context_locked_test
- \core\event_deprecated_test => \core\event\deprecated_test
- \core\event_grade_deleted_test => \core\event\grade_deleted_test
- \core\event_profile_field_test => \core\event\profile_field_test
- \core\event_unknown_logged_test => \core\event\unknown_logged_test
- \core\event_user_graded_test => \core\event\user_graded_test
- \core\event_user_password_updated_test => \core\event\user_password_updated_test
- The following output tests have been moved within level2 directory:
- \core\mustache_template_finder_test => \core\output\mustache_template_finder_test
- \core\mustache_template_source_loader_test => \core\output\mustache_template_source_loader_test
- \core\output_mustache_helper_collection_test => \core\output\mustache_helper_collection_test
- The following tests have been moved to their correct tests directories:
- lib/tests/time_splittings_test.php => analytics/tests/time_splittings_test.php
- All the classes and tests under lib/filebrowser and lib/filestorage
belong to core, not to core_files. Some day we should move
them to their correct subsystem.
- All the classes and tests under lib/grade belong to core, not
to core_grades. Some day we should move them to their correct
subsystem.
- The core_grades_external class and its \core\grades_external_test
unit test should belong to the grades subsystem or, alternatively,
to \core\external, they both should be moved together.
- The core_grading_external class and its \core\grading_external_test
unit test should belong to the grading subsystem or, alternatively,
to \core\external, they both should be moved together.
- The \core\message\message and \core\message\inbound (may be others)
classes, and their associated tests should go to the core_message
subsystem.
- The core_user class, and its associated tests should go to the
core_user subsystem.
- The \core\update namespace is plain wrong (update is not valid API)
and needs action 1) create it or 2) move elsewhere.
This commit changes changes the unit test calculations to
keep track of the time the bucket starts/finishes bursting
and use milliseconds instead of seconds to have a precise
time to sleep.
* As a student I should only be able to access the next question or the current question
* The API should not show more than the current question
* Once the quiz is previewed we can see the question in any order (existing behaviour)
* Related to ticket MDL-71728
Co-authored-by: Rajneel Totaram <rjnlfj@yahoo.com>
Co-authored-by: Tim Hunt <t.j.hunt@open.ac.uk>
Replace steps that manually add H5Pactivity instances via the UI
and use Behat generators. This improves the speed of the Behat test
runs.
Co-authored-by: Simey Lameze <simey@moodle.com>
Editor stylesheets are kept separately to standard plugin stylesheets so
that they can be provided to style individual iframes where loading the
standard Moodle stylesheets would be incorrect.
Unfortunately the editor stylesheet system does not consider that
subplugins may want to add small individual styling to the editor but
this is required in some situations.
The window.tinyMCE.editors API was present in version 3 of TinyMCE, but
is not present in later versions. As a result, this check tries to loop
over a variable which does not exist and throws an error in the process.
We should check that window.tinyMCE *and* window.tinyMCE.editors both
exist before attempting to loop over them.
The changes introduced here are completely safe, just we stop
binding SITEID and, instead, embed it in the SQL.
Why? Because Oracle 21 has started to return non-sense results
when SITEID is bound.
After lots of tests, attempts, debugging... we have been unable
to find any logic to the need of this change and also, have been
unable to reproduce the problem with a standalone script that
pretty much runs the same queries that the ones changed here.
I'm sure that there is something, somewhere, but have failed
to find it, grrr.
Please read MDL-75208 and linked issues to find more information
about this problem, that is one of the biggest mysteries I've
seen recently. Maybe at the end there is a tiny detail that
explains it all, but it's really well hidden.
Adds a warning to the 'name' column of the tables, letting teachers know
that the activity exposed by the method has been deleted and what their
options are.
Moved written table timestamping from query_start() to query_end():
We are adjusting table last written times at the end of transaction.
That does not apply to immediate database writes that are not performed
within transaction. This change is to set last written time after the query
has finished for such writes, rather than before it started. That way
long write operations cannot spill over the latency parameter.
Apply core_form/submit JavaScript code to prevent double submission of login form and guest login button.
Add ID to guest login button to make it addressable.
This is a partial backport of MDL-73270 to ensure that the
check_xmlrpc_usage custom check is present in all the supported
branches and it's applied to all branches able to run php80:
- MOODLE_311_STABLE
- MOODLE_400_STABLE
- master (aka, 4.1 and up)
Note that the whole patch has not been backported, only the
environmental check (xml file and check implementation).
Note that in 311_STABLE the patch is different from the 400 and master
fixes. The API for grade_floatval() was only changed for 4.0 and up.
Hence, for this branch we are just:
- Changing the form definition to use proper "float" element (instead
of text + raw.
- Backporting the behat tests to verify that everything continues
working ok.
Also, in 311_STABLE, the workshop module did not have the grade
to pass completion condition, so we aren't modifying anything
in that activity for that branch.
- Apply the .alert-dismissible class for notification alerts with
close button to fix its positioning. As an added bonus, the
.alert-dismissible class also enlarges the clickable area of the
close button which is great for accessibility.
- Improve example context for the notification alerts templates.
This patch modifies the way copy data is shared in order to mitigate potential race conditions
and ensure that the serialised controller stored in the DB is always in a valid state.
The restore controller is now considered the "source of truth" for all information about the
copy operation. Backup controllers can no longer contain information about course copies.
As copy creation is not atomic, it is still possible for copy controllers to become orphaned or
exist in an invalid state. To mitigate this the backup cleanup task has been modified to call
a new helper method copy_helper::cleanup_orphaned_copy_controllers.
Summary of changes in this patch:
- Copy data must now be passed through the restore controller's constructor
- base_controller::get_copy has been deprecated in favour of restore_controller::get_copy
- base_controller::set_copy has been deprecated without replacement
- core_backup\copy\copy has been deprecated, use copy_helper.class.php's copy_helper instead
- backup_cleanup_task will now clean up orphaned controllers from copy operations that went awry
Thanks to Peter Burnett for assiting with testing this patch.
We are disabling the -v (verbose) option, used by default. Still,
now we support a secret (that can be added to everyone's repository)
to accept any PHPUnit's command line options and run the tests with
them.
Some examples:
phpunit_options = -v (to keep the verbose option enabled)
phpunit_options = --testdox (to print information about every test...)
... (basically anything supported by the CLI)
Also, we are raising here the environment from Ubuntu 18.04 to 20.04
(self tested by GHA, heh)
* Darken the score value using the green theme colour
* Use a light green theme colour for selection backgrounds
* Unify rubric styles related to background and text colour within the
criteria.
* When adding the block_site_main_menu on every page, the activity chooser generated
for this entire page is set with courseid = 1. So any activity is then added to the
home page instead of a course.
It has been detected that the flatten_dependencies_array() was fragile
and leading to wrong results when some incorrect data was passed to it.
This includes:
- Missing elements.
- Null dependencies.
- Non array dependencies.
While the existing behaviour (testing-wise) has been preserved, now the
situations above are better controlled and the function ignores all
those incorrect cases that shouldn't happen ever.
That implies that a good number of notices/warnings/errors aren't
happening anymore. That was impacting both results (when the problems
were only notices and warnings) and execution (when the problems
were errors).
Covered with tests.
Reverting the addition of a Boostrap tooltip on the alert notification's
close button. It introduced a bug where the tooltip remains after
dismissing the notification alert. We can manually toggle the state of
the tooltip via JS, but I don't think it's worth the effort. The tooltip
is meant as an added bonus for sighted users to see what the close
button is about.
Use #region-main-box and not .mform in the selector, otherwise it may
return the wrong mform, such as global search, if enabled. This is the
same way the form is selected in showMultipleSummaryAndHideForm().
Adjusted the condition while reading the streams
in chunks. Read till the end to make sure we have got all
the data. Else we might not get all the data to show
in file picker.
Ensure the saved values for the hidden and hidden until fields are
populated when editing a grade category. Fix behaviour on enabling
the hidden until field.
Re-enable tests commented out in eba1d32c.
As well as storing 1/0 to indicate hidden state, they are also used
to store a timestamp indicating hidden until date. Increasing field
size allows these values to be stored without triggering exceptions.
This adds a condition to check if Online text is used
for the submission. The functions file_postupdate_standard_editor
and file_postupdate_standard_filemanager needs the property
content_editor of the formdata.
* Make the ordering of the message and the close button logical on
the DOM.
* Hide the times HTML entity
* Add sr-only label on the close button for non-sighted users
* Add tooltip on the close button sighted users
In MDL-73915 the nodejs version was switched to lts/gallium but the
travis integration was not updated.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
* No need to enclose the title of the quiz navigation fake block in
a span tag. The block title also gets appended to the skip link which
means that a block title with a span tag assigned with an identifier
can cause duplicate ID errors on the DOM.
When the user is logged in successfully then reset the value
of login_failed_count_since_success to zero, if the value of
login_failed_count_since_success is greater than zero.
Prior to this change, the testcase would assert ordering based on
identical fields which could lead to random failures. Set distinct
next start time of course backup, and also remove random ordering
by sorting on defined course order (all other fields being equal).
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- Moved to the level2 "privacy" namespace:
- \mod_assign\privacy\feedback_legacy_polyfill_test
- \mod_assign\privacy\submission_legacy_polyfill_test
- Moved to the level2 "task" namespace:
- \core_message\task\migrate_message_data_test
- \ltiservice_gradebookservices\task\cleanup_test
- \message_email\task\send_email_test
- \mod_lti\task\clean_access_tokens_test
- \mod_workshop\task\cron_task_test
- Moved to the level2 "event" namespace:
- \core_h5p\event\deleted_test
- \core_h5p\event\viewed_test
- Renamed to a better name:
- backup_forum_activity_task_test.php (missing "task")
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- Some fixtures, initially defined in the test files have been
moved to new files in fixtures subdirectory, leaving the unit
test files clearer:
- moodle2_course_format_test.php
- Rename wrong named test:
- baseoptiogroup_test = baseoptigroup_test
This patch wraps the login_failed_count logic in a resource lock and
forces a user preferences cache reload. Each thread must wait for the
lock and must fetch the current count before incrementing it. This
ensures that login_failed_count is correct across threads and that the
lockout threshold is correctly honoured.
Co-Authored-By: Sujith Haridasan <sujith@moodle.com>
Modified course format options reading and writing to be able to handle Editor elements by enabling them to split array values into
multiple values before inserting into database, and combining multiple values into an array when reading from the database.
Modified backup and restore code to use backup_nested_elements, and to interact directly with the database.
Co-authored-by: Jason den Dulk <jasondendulk@catalyst-au.net>
Co-authored-by: Matthew Hilton <matthewhilton@catalyst-au.net>
Modifies the existing upgrade step to only take into consideration
block instances that have a specific category set prior to determening
if a legacy category is being used and updating them. Instances that do
not have any category set should be ignored.
Upgrade step to identify all existing cases where users have linked
their MoodleNet profile on the site and remove the related data from
the database. Due to some recent changes on the MoodleNet platform,
this data is now irrelevant and can no longer be used to authenticate
on the MoodleNet platform.
Updates the existing language strings in order to provide more accurate
instructions on how users can link existing their MoodleNet profile on
the site.
Makes sure that the correct file extension is attributed when importing
a file using the Google Drive repository based on the set configuration
values for the default import format settings.
Do not show summary and scales for the survey with critical
incidents. So critical incidents survey would have questions
and participants only under response reports.
Also corrected the heading for the Questions page.
This is a BEHAT_RUNNING only hack, so it doesn't affect normal
operations at all. What it achieves is to reduce the number of
strings loaded when customising a lang pack from current 31K ones
to the just needed ones for Behat testing purposes.
That way we avoid the random failures that are happening more and
more (while the "en" lang pack grows) and also save some precious
minutes in every behat run.
Also, unrelated, a couple of tiny changes:
- Modified constant (that was really outdated) to current number
of lang strings (so the progress bar behaves better).
- Remove an elseif occurrence because it was hurting my eyes.
Before this change we were setting the month day to 1 unconditionally
and that was working ok for enabled by default elements.
But in a number or cases we were doing that against disabled elements
so the day reset was not happening. Now we wait for the element to
be enabled and only then proceed with the day reset.
Backed with tests using disabled form elements, so we have now
both enabled and disabled fields covered.
There is an edge case in the conversion from old collapse/expand
settings (stored as user preference) to the new ones that can lead
to notices / warnings / errors.
Before PHP 8, they weren't being detected, but now they are.
Basically, if some of the expected components of the preferences
are missing, then some array (array_diff(), array_intersect()..)
operations now are failing, because now those parameter functions
do require to be arrays (so empty strings or nulls aren't allowed).
Also, there were some cases where the old preference was not being
removed ever, so its removal has been moved to a common place at
the end of the function, where it's always evaluated in case
the preference still exists but is empty of contents.
I've tried to cover all the cases I've been able to imagine with
unit tests (empty information, some component not set, all components
not set...), have moved it to a new unit test because the one
already performing some tests with those old preferences was
already too long to add more cases.
MDL-66920 added this field but didn't set defaults for upgrading tool
types. This patch fixes that for any 1.3 tools which don't yet have a
value for this field.
* Fetch only the fields required by the function.
* Quick validation of the passed forum parameter. If the forum parameter
is an object, it should have the forum ID and the course ID.
* Default return when there's no need to show a warning yet.
* Return early if the forum's blockafter or blockperiod attributes
are empty.
* If $cm is not passed in forum_check_throttling(), try to fetch it
using get_fast_modinfo() which avoids DB reads. Fetch it via
get_coursemodule_from_instance() as a last resort (though it's unlikely
to happen).
AdoDB SQL*Server drivers use to modify some error and
logging preferences. As far as the connection is shared
with the main one, that implies that those settings
remain for the whole remaining execution.
We don't want his in tests, because can affect to other
tests at distance.
We are fixing this here, only for tests, because it
doesn't affect normal requests and because, once we
remove AdoDB this won't be needed anymore.
Sometimes, in order to provide a cross-db behaviour of unique indexes
mixing null and not null columns, we create, under the hood, some
function based indexes.
When that happens get_indexes() is returning the name of the
expression objects used to calculate that function index. But we need
the original column names to be able to compare indexes and get
column dependencies properly.
So, this patch just looks, when the index is unique and function based,
to the expressions (pretty standard CASE statements) and gets the
original column name from it.
Covered with tests.
Use PARAM_ALPHANUMEXT for $areaid as areaid can also be
say mod_h5pactivity. When PARAM_ALPHAEXT is used it would
become mod_hpactivity and would cause exception in code flow.
When a student post exceeds the posting threshold
block then 'Add discussion topic' button should not be shown.
Also the student cannot reply to a topic if this is exceeded.
This commit is UNRELATED to the MDL-74184, but we need the
@skip_interim tag removed to a couple of scenarios that were
causing lots of noise and failures @ci infrastructure.
MDL-74265 was created to deal with this issue but these jobs
are now passing, and nothing has been changed...
So we are using MDL-74184 (this extra commit) to remove that
tag (sorry for the noise).
The "Save changes" button in the SCORM package falls out from the
viewport and that seems to be a problem for Firefox, not being able
to auto-scroll to it before clicking.
So we are just making the window 200 points taller and that
makes the button to be visible (within the viewport) and clickable.
Also, change the way we "wait" at the end of the scenario and,
instead of reloading, we re-visit one of the contents to allow
completion to be ready for final assertions.
In very slow environments viewing the lesson page for a student takes
more than 1 second and, because of this, the "Spend at least 1 sec"
completion condition is considered done.
So, in order to alleviate this problem, that leads to random failures
in CIs with slow databases, we have changed the condition to be 3
seconds instead.
That seems to do the work in super-slow environments, with 100
executions passing ok (when previously the scenario was failing 33%
of times).
The only cost is that now the behat run will be 2 seconds slower (there
is a wait changed from 2 to 4 seconds), but that's an acceptable
price to pay to get the scenarios consistent passing.
See the issue for more details.
This just applies two "tricks" to make the test to behave better:
- Ensuring that the "wait until the page is loaded" happens in the scorm frame.
- Ensuring that the last item in the package gets some extra time
by reloading the page completely at the end.
With them, a testing environment with a 100% of failures has become 100%
passing (25 repetitions). That should be good enough to reduce the
impact of that failure everywhere.
See the issue for more details.
This commit is UNRELATED to the MDL-74116, but we need the
@skip_interim tag added to a couple of scenarios that are
causing lots of noise and failures @CI infrastructure.
MDL-74265 has been created to deal with this issue.
So we are using MDL-74116 (this extra commit) to add that tag
(sorry for the noise).
If updating users during upload along with the option to fill in
missing fields from file content, then we should exclude the auth
column because it will always be present for a user.
The previous iteration only calculated it's zIndex value on
initial load. This meant that any nodes added subsequent to this
would not be taken into account (e.g. modal forms).
- New course_modinfo::get_array_of_activities() function was created to reduce
the unnecessary DB call
- New course_modinfo::purge_course_cache function was created to purge the cache of a given course
When we change the position of two sections, we just need to update
the position of the affected sections, not all the sections in the course.
This will improve the performance since the system only executes the queries to affected sections.
Also, the system only clears the cache for affected sections, not all the sections in the course.
The course view is updated to display the events that match
the filter condition, no matter the course start date is in the future
or the course end date is in the past
An icon has been added near the 'Help and documentation' link in the
site footer When the site admin setting 'Open in new window' (doctonewwindow)
is enabled, for consistency with other similar links like 'Give feedback
about this software'.
Apart from that, now the link will be opened using the standard target
'_blank' instead of using the new window Javascript, to let users decide
whether the new page should be opened in a new tab or a new window.
When "Path to PHP CLI" is not defined, an exception with the message
in cannotfindthepathtothecli should be displayed, and the page should
be redirected to System paths settings page.
Apart from that, this patch also replaced core_task to tool_task,
because this message wasn't traslated properly.
Allow email account verification to be disabled for any Oauth2 provider.
Also add clear indications to administrators of the danger of doing do,
this is done by an additional form checkbox.
This patch also reverts MDL-66598
For PHP 8 and up, the 'mbstring.func_overload' doesn't exist
anymore, so skip any check/condition based on it.
The patch has been done that way, so it's easier to remove
the whole block of code once PHP 8.0 becomes our min supported
PHP version.
The change is self-coveres (changes the external and its own tests).
The 'mbstring.func_overload' php init setting was removed for
php80 (it was deprecated since php72). So it won't evaluate to
true ever, so the whole block can be put under php version condition.
Note that this is already fixed upsteam, for commit:
https://github.com/bobthecow/mustache.php/commit/e7165a33b282ab4d20b3863825caadb46313d62b
And it has been released with version 2.14.1 of the library, so, once
we upgrade to it, the fix will be incorporated.
For a better PHP8 compatibility this commit replaces the deprecated $php_errormsg
with error_get_last().
The PHP 8.0 migration guide (https://www.php.net/manual/de/migration80.incompatible.php) says:
The track_errors ini directive has been removed.
This means that php_errormsg is no longer available.
The error_get_last() function may be used instead.
Also the documentation on $php_errormsg (https://www.php.net/manual/en/reserved.variables.phperrormsg.php) says:
This feature has been DEPRECATED as of PHP 7.2.0.
Relying on this feature is highly discouraged.
Use error_get_last() instead.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
The PHP_CodeSniffer @codingStandardsIgnore annotations are deprecated
and, since version 3.x, the new // phpcs:ignore comments should be used
instead.
This commits just reviews all the uses in core, replacing them for
the better new candidate, or removing when no longer needed.
The OBv2.0 specification includes a field "Criteria" for
BadgeClass. Until now, this field was filled using the
URL of the badge assertion, but that is causing some issues
in Badgr because it linked to the badge assertion of the
first user sending this badge to the Badgr backpack (so then,
the following users linked to the first user assertion page
too).
This patch adds a new page, badgeclass.php which will be
used from now to display any badge information which is
not related to any assertion (like happens with the criteria
in BadgeClass).
This is a backport from MDL-72141.
Apart from reorganising the information displayed in the external badges
page, this information has been moved to reuse the template created in
MDL-71937, to make it easier to modify and maintain.
This is a backport of MDL-71937.
Apart from reorganising the information displayed in the badges page,
this information has been moved to a template in order to make it
easier to modify and maintain.
Various tests have been added to cover the points fixed in the previous
commit, using a localised decimal separator (#). Using the same
numbering, here there are the details:
A. grade/report/singleview/tests/behat/singleview.feature
grade/tests/behat/grade_category_validation.feature
B. grade/tests/behat/grade_override_letter.feature
C. (done within various scenarios)
D. grade/report/singleview/tests/behat/singleview.feature
E. admin/tool/behat/tests/behat_form_text_test.php
F. grade/tests/behat/grade_to_pass.feature
mod/lesson/tests/behat/lesson_informations_at_end.feature
G. quiz/tests/behat/preview.feature
H. mod/quiz/tests/behat/info_page.feature
I. question/type/ddimageortext/tests/behat/edit.feature
It has been detected, thanks to php80 specially, that there are
various places in core where support for localised floats is
missing. Before php80, some locale-dependent conversions were
performed by PHP, allowing things to work. But with php80 all
those comparisons are now locale-independent. See:
https://wiki.php.net/rfc/locale_independent_float_to_string
That implies that we now need to, always, unformat floats to
be internally the correct (decimal point as separator) in
order to compare it.
While this was visited in the php80 epic (MDL-70745), nothing
was found, all automated tests were passing ok. Problem is that
we run behat tests with en-AU laguage that has the decimal point
separator.
So, in this issue we are fixing all the problems detected by
running those Behat tests using localised (comma) decimal
separator.
Note that there may be other places still causing problems, but
it's really hard to find them programmatically, so we'll have to
wait for real use reports / issues and go fixing them while they
happen.
Back to this commit, this is the list of changes performed (note that
in the next commit, we'll be adding scenarios explicitly using
a localised decimal separator to ensure that they work ok).
A. Changes to various grade forms to ensure that, on their validation
floats are unformatted properly. Also, changed the corresponding
form element from current text/PARAM_RAW to proper float ones that
take care of the conversion in a number of places (but when disabled,
that's the reason we still have to unformat in validation.
This includes the following forms:
- edit_category_form
- edit_item_form
(this is the original problem reported that cause all the research
to be performed against full behat runs)
B. Changes to edit_letter_form, so it uses a proper PARAM_LOCALISEDFLOAT
(note this is the type of change that surely should be used for all
the rest of /grade/edit/tree form, including those in the previous
point).
C. Changes to the grade_item behat generator, so it's able to work with
localised floats, un-formatting them when needed.
At lib/behat/classes/behat_core_generator.php
D. Fix problem passing localised floats to scales, not displaying
properly. At grade/report/singleview/classes/local/ui/finalgrade.php
E. Change the behat text matcher in order to allow comparison of
localised floats when they are the current ones. Before this change
the matches was using soft/lazy comparison, so '50' and '50.0000'
match. Now, when the comma (for example) is used (and only then),
'50' and '50,000' will also match. This comparison is in use in a
bunch of tests and makes sense to make it localisation-aware.
At grade/report/singleview/classes/local/ui/finalgrade.php
F. Fix a couple of number_format() uses in lesson, because they are
not localised-aware. Switched to format_float(). At mod/lesson/locallib.php
G. Change the quiz_contains_the_following_questions() step to accept
localised maxmark expectations. At mod/quiz/tests/behat/behat_mod_quiz.php
H. Change the quiz generator so it accepts localised gradepass.
At mod/quiz/tests/generator/lib.php
I. Change the edit question form to show proper localised penalties,
previously it was always showing point-decimal ones. Of course,
leaving the values of the select element unmodified (internal floats).
Related, also change a couple of tests to, instead of try to match the
value (always internal floats), match the description (now localised),
so we can test them with different separators. At:
- question/type/ddimageortext/tests/behat/backup_and_restore.feature
- question/type/ddmarker/tests/behat/backup_and_restore.feature
- question/type/edit_question_form.php
On some larger sites, processing the navigation with format_string will
habitually load all the contexts for navigation which can take 400+ DB
queries. Explicitly tying all those format_string calls to the system
context reduces this overhead to a single DB query that probably has
already been run on the page previously.
Co-authored-by: Peter Burnett <peterburnett@catalyst-au.net>
Prior to this change, the tool_task_cronrunning check had very tight
limits for checking the last cron interval and comparing it to the
expected frequency. When a cron job should take 1:00 min and the last
run was 1:02 min it complained that there was 1:02 between the last two
runs of the cron maintenance script and it should run every 1:00.
This change makes the check a bit more relaxed and adds an additional
minute on top of the expectedfrequency to give the cron job some time.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
To all the walkthrough, questiontype, question, upgradelibnewqe,
behaviourtype_test and edit_form testcase classes::
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- When belonging to other component and being valid api:
- form related tests have been moved to tests/form subdir.
Applied the following changes to various testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- When belonging to other components and being valid api:
- analytics related tests have been moved to tests/analytics subdir.
- backup & restore related tests have been moved to tests/backup subdir.
- events related tests have been moved to tests/event subdir.
- privacy related tests have been moved to tests/privacy.
- task related tests have been moved to tests/task subdir.
- Some simple renames, not including the component part anymore (not
needed now that they are namespaced):
- some xxxlib_test.php have been renamed lib_test.php
(when they where testing the corresponding lib.php).
- cache stores tests have been all renamed store_test, originally
each one had its own name (file_test, apcu_test, redis_test...)
- assign feedback tests have been all renamed feedback_test, originally
each one had its own name (file_test, editpdf_test...)
When questions are deleted, we now check if the parent is in use before deletion.
Prior to this, it would be possible for multianswer questions to reference
questions that have been deleted. This results in fatal errors when the quiz
is viewed.
This patch uses a dummy 'subquestion_replacement' to handle this case and
display some information to the end user about what has happened so that
they may take action to repair the corrupted question.
As a result of the bug described above, the sequence column of
mdl_question_multianswer can contiain references to questions that no
longer exist, and these IDs can make their way in to backups.
When this happens, the backups cannot be restored. To avoid this,
this patch skips trying to restore those questions that reference
question IDs that no longer exist (as there is no way to recover them).
* As per WAI ARIA Authoring Practices 1.1 guidelines for menus,
menu items should have a tab index of -1. Navigation between menu items
is done via arrow keys.
See https://www.w3.org/TR/wai-aria-practices-1.1/#menu
* Move the focus to the last menu item when the menu is displayed by
pressing the Up arrow key.
* When the menu is open and Tab/Shift-Tab is pressed, focus on the
next/previous focusable element on the DOM instead of focusing back on
the menu trigger.
* Combine event handling for the dropdown menu trigger.
Unfortunately the babel minify-mangle plugin seems to be abandoned and
in certain circumstances can be very buggy. The only safe options are to
disable it, or to switch to a different minification library.
Not minifying our javascript is not ideal, so this commit updates the
javascript tasks to use a rollup, combined with babel, and terser.
Babel still converts code from ES/UMD/AMD to AMD modules with the
relevant browser support, whilst terser minifies the code.
The rollup bundler handles tracking and creation of sourcemaps, and
supports better parallelisation of the tasks.
Since the upgrade to Node LTS/Gallium requires an upgrade to @babel/core
and eslint, which change the built files anyway, this seems like the
ideal time to make this change.
This commit removes the moodle-behat-extension from the list of
third-party libraries.
In order to do this properly, all pending coding-style issues are fixed.
In addition the following niceties have also been included:
* convert long form array() syntax to short form
* alphabetically sort class imports (use)
* remove unused class imports
* correct documentation
* Normalise whitespace in concat operators
* Remove unnecessary file docblocks
This was leading to exception:
Class 'core\message\coding_exception' not found
Surely we weren't facing this before MDL-67853, but it seems that
now we are hitting it sometimes (being investigated @ MDL-73747).
So, just ensure the proper \coding_exception is used.
If a component has no available notification types (because they're
disabled), then it (the component) should be hidden entirely from
the user notification preferences page.
Summary:
- With PHP7, libxml_disable_entity_loader() was making those
flags futile, and the DTDs were not being requested ever.
- With PHP8, the flags get precedence and requests for the
DTD have started to happen.
- Those requests are not served by W3C (1 minute time-out).
- Those DTDs aren't used by anything in code (no entity
replacement, no validation...)
Hence:
- Remove the flags so it will work the same in PHP7 and PHP8.
- Just to double ensure we don't want any remote loading
to happen ever, add explicitly the LIBXML_NONET flag.
Because of the way in which we check for exceptions, we always show the
following string during behat pretty output:
behat_hooks::i_look_for_exceptions()
We can easily modify this to only show this value when the
i_look_for_exceptions() test itself fails, but otherwise to show the
actual step that is run.
There is an issue with the current Badgr manifest because the image
URL contains double slash which is causing an error when storing it
in the database (https://api.test.badgr.com/static//images/logo.png).
This issue parses image URL and removes multiple slashes in URL.
This was causing problems when sections or activity names where
set to be empty or whitespace-only, with the PARAM_CLEANHTML
leading to exception.
Now they are supported and handled like (0xc2a0) to
allow the process to continue.
Added a few extra tests to confirm the behaviour.
Before this commit, is_empty() was being applied before returning
the mock response. But we want to be able to mock the empty response
for some tests, hence moving the condition to null/isset, that is
the value that array_pop() returns where there aren't more elements
in the array.
With that change performed, we can test lti_load_cartridge() with
empty responses, hence adding a new test for that.
Sometimes (detected with Windows, when running @ GHA), both the
response and the error of a curl request to non-existing URL
returns the empty string.
In that case, we cannot call to DOMDocument::loadXML() because the
1st param cannot be empty. So here, whenever that happens, we are
throwing the moodle_exception earlier, instead of waiting for the
XML errors to be processed later.
When the oAuth2 issuer hasn't any userinfo endpoint, a call to
$this->get(false) was done, which was returning "The URL is
blocked".
This is a regression from MDL-70649, which added some cURL security
checks.
Sometimes, for the question that has a lot of input groups and unlimited draggable items,
the 'clone' process takes longer than usual, so the questionManager.init() method
will not add the eventHandler for the cloned drag.
We should make sure to bind the eventHandler for the cloned too
Note that, instead of going one by one to them from the XMLDB Editor
(load, makes some tiny change, save, revert the change, save), the
report includes a commented line that, once uncommented, does regenerate
the file while reporting.
I've left it there as a commodity for the developer.
People uses to edit the install.xml files manually, here and there. That
uses to come with small mistakes, like wrong white-space indentation,
some attributes out of order...
While none of them are critical, they add a lot of noise when somebody
is correctly editing a file with the XMLDB Editor because it, correctly,
overwrites the whole file and then diffs show a lot of unrelated changes.
So, this report will inform us about any file needing to be regenerated
because it has been manually edited (in a different way than the way
the editor has).
Note that the report is very basic, with minimal ouput, manually
generating the HTML, like the rest of the XMLDB Editor actions do. We
are not using renderers neither templates here.
Also note that it includes a commented line of code that, once
uncommented, enables the report to, also, fix the wrong files. Useful
for developers.
The icon margin added in MDL-73543 is causing multiple alignment issues. The alignment issue it needed to fix is done for the recently accessed course block
only.
Prior to this patch, the code was fetching all completion data for all
activities in a course, even when the activity was not requested. This
leads to recursion issues as the data has not been added to the cache
before this operation occurs.
To handle this situation, only the requested CM is fetched in full, and
a boolean flag is used to store whether the full data has been fetched.
When returning a partially fetched value from the cache, the flag is
used to determine whether more data must be fetched, and the cache
updated.
The flag is filtered out before the value is returned.
Note: Many of the tests were updated as these were inspecting private
features of the API which should not really be tested.
And replace them by phpunit @covers annotations including
all the classes declared or extended in the removed members.
It's not awesome coverage, but better than nothing.
It's been a mistake to make every Moodle site out there to check for
available update every two hours, on the hour. We are experiencing
server load peaks that are harder and harder to deal with.
The patch addresses this by (1) checking for available updates just once
a day and (2) distributing the checking times randomly around the clock.
All restore_date_test, api_test, rule_test, plugin_test,
manager_test, helper_test testcase classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Remove file phpdoc block
- Remove MOODLE_INTERNAL if not needed.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to:
- All restore_date_test cases have been put under xxx\backup
level 2 (valid API) namespace.
Since php80 the 5th parameter (cypher) in the following functions:
- openssl_seal
- openssl_open
is mandatory, hence:
- adding it.
- with value 'RC4' to keep BC.
The course_get_user_navigation_options() is called on each request
from the add_course_essentials() method, to initialise the basic
navigation options that are available or not for the current user.
This information is not always displayed in boost (for
instance, in a module page), but it might be required by some blocks,
like Navigation.
However, this block is hidden by default in boost since MDL-73347,
so at any point we should check if it's possible to change the way
this navigations options is loaded, to only get this information
when it's strictly required.
After cbf9dfb the CURLOPT_FILE no longer behaves as expected. All
redirect responses are appended to the same stream resource. This fix
reverts back to the old behavior by setting the stream pointer back to
the beginning for each subsequent redirect.
If there are no course badges, students shouldn't have a link to a
page saying there are no badges available.
This patch is for displaying the Badges in the secondary navigation
only if the user can manage badges or there is, at least, one
badge available to the current user.
All externallib_test, generator_test and filter_test classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to tests under testing/tests:
1) The core_testing component doesn't exist.
2) But testing/tests are allowed because there is a suite pointing to it (phpunit.xml).
3) So, the only possible namespace for them is "core".
4) And to avoid problems with other core testcases (under lib/tests)
they have been renamed to have testing_xxxx as prefix.
Finally, also modified calendar/tests/event/events_test.php because it uses
some renamed (core_calendar_externallib_testcase => \core_calendar\externallib_test)
classes.
This commit removes code that only was being executed by php < 73
and it's 100% safe to do so because Moodle 3.11 and up require
php 73, hence it was not executed ever.
Removed code includes:
- ldap_control_paged_result and ldap_control_paged_result_response
(that were deprecated in php 73 and have been removed in php 80).
- conditional code in the session manager, where some hacks were
needed for php < 73. Note that this removes the private function
append_samesite_cookie_attribute() completely because it was
doinf nothing (first line was returning for php < 73).
- Also removed the old session.hash_function ini setting because
it was removed in php 71.
Kept code includes:
- The environmental check_igbinary322_version test has not been
removed because it doesn't hurt (always returns "ok" for php 73
sites) and doing it would involve to backport the environment.xml
file to 39 and 310. Instead, a note has been added to MDL-71747
in order to get rid of that check for 4.1 and up.
So we are putting them under a standard PHP_MAJOR_VERSION < 8 condition.
Also, added TODO comments to remember to delete that block of code when
php80 becomes the minimum required versions in the future (Moodle 4.3?).
When possible, an issue has been created upstream, else a comment in
readme_moodle files has been added.
Finally, when the keys being freed were class or object attributes, also
nullify them (but when the calls were part of destructor methods).
The menubar role defined in the "menubar d-flex" element is not
required and it's causing the following error "Required ARIA children
role not present: group, menuitemradio, menuitem, menuitemcheckbox".
All events_test, external_test and search_test classes:
- Namespaced with component[\level2-API]
- Moved to level2-API subdirectory when required.
- Fixed incorrect use statements with leading backslash.
- Changed code to point to global scope when needed.
- Fix some relative paths and comments here and there.
- All them passing individually.
- Complete runs passing too.
Special mention to lib/external/tests/external_test.php:
- external is not a valid component
- hence, its only valid namespace is "core"
- also, it's testing lib/external/externallib.php
- hence, the file (and class) have been renamed to external_externallib_test.php
(to avoid conflicts with other external_test that may exist in core)
It never has been used in core since it was introduced in 2011
by MDL-22112, because the ouwiki_diff_html() is used unconditionally
for all wiki formats (html, creole or nwiki).
So, removing it to avoid any current (php80) or future incidence
with it.
Move a module to another section needs to update its visibility after
placing it into the new section. Otherwise an invalid database state
prevents the system to move any module with a completion rule
involving grades.
This matches it's type in the settings pages, and also the type
of the site name used as the default for the setting. It's content
will now be automatically filtered by the external exporter class.
They were deprecated in PHP 7.4 (warning) and have been removed in
PHP 8.0 (error). Plus they were returning false since ages ago (PHP
5.x).
Hence, adding conditional check to see if they are available (h5p
editor) or removing the unreachable code-block when they evaluated
to true.
Also, modified the corresponding moodle_readme files to document
the change.
All lib_test and locallib_test classes:
- Namespaced with component (and API whenever makes sense).
- Fixed incorrect use statements with leading backslash.
- Changed code to point to global scope when needed or add new uses.
- All them passing individually.
- Complete runs passing too.
Special mention to tests under login/tests:
1) The core_login component doesn't exist.
2) But login/tests are allowed because there is a suite pointing to it (phpunit.xml).
3) So, the only possible namespace for them is "core".
4) And to avoid problems with other core testcases (under lib/tests)
they have been renamed to have login_xxxx as prefix.
Custom profile fields allow for shortnames containing mixed character
casing, which should also be supported within the user fields API
when retrieving said custom fields.
Given that the DML libs convert all selected columns to lowercased
equivalent, we need to do the same when requesting custom fields to
ensure calling code doesn't try to access `->profile_field_Foo`
property when the DB returns `->profile_field_foo` instead.
When there is no group but a default groupmode is set, the group
configuration is provided but the group data is not present. This
resulted in a JS error and an invalid filter line was left in place.
This change conditionally checks the data-filter-type-class value of the
filterDataNode so that no JS error is produced if the filterDataNode
does not exist and therefore removes the empty filter row.
This was detected when working in the renaming of test names:
- Fix file name, it was incorrectly named .test.php, now _test.php.
- Add namespace and rename testcase name to match file name.
- Fix php80 deprecation message about optional and required params
order.
As far as now all them have correct privacy level2 namespace:
- Move them to "privacy" subdir.
- Rename the files to "provider_test.php", this includes old
privacy_test.php and privacy_provider_test.php files
- Rename the testcase to provider_test too (to match file name)
Also, change some relative paths and comments to point to new
locations.
All privacy_test and privacy_provider_test classes:
- Namespaced with component\privacy.
- Fixed incorrect use statements with leading backslash.
- Changed code to point to global scope when needed.
- Renamed a few files to make all be privacy_test or privacy_provider_test.php
- All them passing individually.
- Complete runs passing too.
While testing the changed unit tests one by one, the
payment/tests/helper_test.php was failing because of a mismatch
between the dataProvider data and the param type.
That made me think why that error was not popping in complete runs
and the answer is that the new (3.10) subsystem is not in the
phpunit.xml.dist file.
Hence, adding it and fixing the dataProvider mismatch.
Finally, it has been detected that some old ICU versions (< 62)
have an incorrect behaviour, not adding the required non-breaking
white-space between the currency abbreviation and the value. So
we are skipping some payment tests if that's found.
The mod_lti css rule .path-mod-lti .message conflicts with text
alignment in chats in message an can be removed. The last occurrence
of the css class "message" was deleted from the mod_lti code in 2011
with 036e84c389 (see MDL-30354).
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
Within the renamed classes in the previous commit, there were a
few incorrect namespace, this commit fixes all them to be
correct component[\api] ones.
Also fix a few incorrect leading backslashes in namespace/use
statements. And tests extending renamed tests.
These are the only cases 100% safe to apply the renaming of the
testcase class names to match the file names.
All other cases are not safe, because they are missing namespace
and may enter into name conflicts. Adding namespaces is not as
simple as imagined because it implies to, also, add a good number
of modifications to core. See the issue for more details.
Prevents DML debugging from duplicate values in first field, also
fixes `context` class namespace and removes undefined `localrole`
property of $role object.
When session lock debugging and read only sessions deubgging are both
enabled, session lock debugging becomes moot. This patch causes
the session lock deubgging code to exit early if read only sessions
debugging is enabled.
The random retry delay for redis session cache was calculated as
rand(100000, 500000) giving an effective retry delay of 100 seconds
to 500 seconds. That's off by a factor of a thousand! Using Redis as a
session cache and when the connection hangs, you can get random
"cannot obtain session lock" errors because it's waiting up to
500 seconds (or about 8.33 minutes) for a Redis connection.
This sets the delay to the originally intended 100ms to 500ms.
(see MDL-59866).
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
On account creation by Oauth2 all configured mapped user
fields are now saved in the user profile on user creation.
Co-Authored-By: SKOLL Learning Technologies
The previous iteration using the behat_form_field class directly
didn't explicitly require the file that defines this class, and in
Goutte runs this led to fatal errors (class not found).
In some cases, image might have more fields rather than the URL.
This patch is for taking them into account and displaying the image
and also these information properly.
PHPUnit test dataProvider methods should not modify global state,
such as initializing $OUTPUT, as they can cause failures in
unrelated tests at a distance.
When a third-party AMD module is provided with an included name we must
replace it with a name that we are able to use.
The structure of an AMD module define function is:
[name], [dependencies], [callback]
String, Array, Function
Each of these is (strangely) optional.
We know that the name is the only String and will always be the first,
and we can use AST to remove a provided name before adding the
Moodle-specific name.
This should be a safe change and not lead to any changes in built code,
but will make it easier to include third-party code.
An additional logging line is also included to inform developers that a
change was made.
Quiz completion settings are only validated if they are unlocked.
Undefined behaviour in the forms API results in the completion
settings 'require passing grade' and 'completion attempts exhausted'
not being sent to the process options function if completion
settings are locked. This resulted in the completion attempts
exhausted setting being disabled whenever the quiz was saved
as it did not detect the require passing grade setting.
This changes the behaviour so that validation is only performed
if the completion settings are unlocked.
A behat test is included to ensure the setting does not change.
The format_string() function does half html escaping. It escapes < and >
characters, but does not escape ". Therefore, it put us in a situation
where neither {{}}, nor {{{}}} are suitable in templates. If we use {{}}
for an attribute (like title or aria-label), then < and > characters
will be double escaped. On the other hand, if we use {{{}}} there, a
double quote character will break HTML when it is used in an attribute.
Therefore, neither {{}}, nore {{{}}} are usable in html attributes.
Moreover, The title attribute here was redundant because it had the same
value as the link's text.
The new get_safe_orderby() and get_safe_orderby_multiple() methods
provide a centralised safe way for user submitted sorting values to be
incorporated into SQL ORDER BY. They do this by removing the need for
user submitted data to pass in any SQL and not allowing arbitrary
column values, instead using string keys which map to a predefined
list of allowed sortable columns.
Checks if a form element has attributes before
trying to get the default value. This fixes
an error when trying to get the default value
for a frozen group which does not have the
attributes array.
* Adjust the margins to accommodate the button outline so that it
doesn't look clipped.
* Inherit the opacity when the button receives focus to improve the
focus outline's contrast.
The problem (coincidence of contexts) was detected when running
oracle phpunit tests, so we skipped the tests as part of MDL-72743.
Now, with the problem fixed (previous commit), we can bring back
that test.
Though it's really rare for this to happen (it only was discovered
when running unit tests with Oracle), it's possible to get
problems restoring courses when context ids in the backup file
do match existing contexts in the restore target site in certain
ways (see the issue for more information).
This change just ensures that every call to the method (that
happens 4 times, for sys, coursecat, course and module levels):
prechek_precheck_qbanks_by_level()
Is processed always for that level. Before the patch it was
possible (hardly but possible) to return question categories
belonging to another level when some qcats contexts were matching
between the backup file and the target site.
- Label the table/list with aria-label
- Hide sort buttons from screen-readers and use aria-sort instead
- Remove the redundant title from decorative thumbnails
Before this change if a new post was made after the digest time but
before todays daily digest had been sent the run time on the digest
would be set to the next day resulting in no digest being sent to
the user.
By adding the server midnight to the custom data on the task we
ensure that in this case a new adhoc task is created for the next
day leaving the current one to get processed.
Server midnight is used so that if the digest time setting is
changed we would not get two tasks queued for the same day.
Before this patch the exported users (to csv file that jmeter consumes)
were all the enrolled users in the test course. And that's ok when the
number of users enrolled match the number of concurrent threads planned
for a jmeter plan (each thread is a user).
But when both numbers don't match, that can lead to the jmeter plan
not behaving as expected, because it iterates over the excessive users
in the file, leading to some users having run 5 loops, others 4...
The only way to make results more consistent is to, always, ensure that
the number of users exported for the plan match the plan size and not
the site size.
And that's what this issue exactly does, restrict the export to the
number of threads that the plan will have. So every user always runs
the very same number of loops.
Ideally any new test should be using a class name matching
the file name, and to use namespace when clearly belonging to
a component (sub-namespace also allowed for better accuracy).
And, also, adding a @covers label to indicate which class
(preferred, method is also possible) is being covered with the test.
This also implements the functionality for SCORM packages, which may
have SVG images embedded within their content which require loading
directly (and are only created by users with appropriate risk flag).
Differences can be checked with git diff --ignore-all-space and
that will show that the only non-whitespace change is in a test
in the git lib/tests/html2text_test.php file. In that case the
whitespace has been replaced by a quoted string (of whitespaces)
keeping the test 100% equivalent and workarrounding the whitespace.
The names and summaries of course elements (i.e. the course
itself, plus sections and modules) included when downloading
it's content are now passed through appropriate methods to
format their values.
In some situations it is useful to modify the requested screensize by a
modifier. This allows for failing tests to be rerun at a different
screensize, where the original screensize triggered a failure for some
size-related reason.
This change adds configuration for the eslint-jsdoc plugin in order to
detect the following errors with jsdoc blocks:
* missing params
* params in incorrect order
* params with no type defined
* properties incorrectly defined
* incorrect use of @package
* incorrect values for @access
In addition, warnings will be thrown where indentation is incorrect.
This is the 3rd round adding the tags to some scenarios
detected to be failing with zerosize problems lately.
Note that @ MDL-71108, the final solution will happen, say
switching to own webdriver till the upstream one is fixed
or whatever. Our @skip_chrome_zerosize solution is just a
workaround to have all those scenarios detected and ignored
by normal runs.
Without modifying any installed package or dependency
because, after trying to do so, the updated stuff is huge
and leads to some JS validation / minimisation problems.
The previous method of hardcoded filepaths, specifically for events
belonging to core, meant that any events belongs to core subsystems
were omitted (core_customfield, core_h5p, core_payment, etc).
For classes with many students, there may be valid cases for having more
than 30 reviews per submission / reviewer.
The patch changes the type of the input field so that it can freely
editable to any integer value.
Prior to this patch the debugging mode (when enabled) would trigger
on everywhere, regardless of whether or not READ_ONLY_SESSION is defined.
This patch modifies that behaviour so that the debugging only kicks in
if READ_ONLY_SESSION is defined and set to true.
The availability condition get_description method is called while
gathering data for the modinfo object. As such it is not safe to
call other functions which might rely on modinfo, such as format_string
(if using filters which access modinfo).
This change provides a mechanism to call format_string later, and also
a general callback to do other stuff later as well if needed. It uses
the same approach already taken to make activity names work correctly
in the availability_condition class.
Course category fields were changed en masse from simple select
elements to autocompletes in 93d72205, but weren't defined as
required fields. This could lead to exceptions and/or unexpected
behaviour if their values were cleared prior to form submission.
More notably PHPUnit to 9.5.9 that comes with some good fixes
and better support for go prospecting php81 (beta) status.
Generated with php73, that is the lower version supported
by this branch, following the instructions @ https://docs.moodle.org/dev/Composer
* New pix icon i/externallink
* New lang string opensinnewwindow
* Use the external link pix icon with "Opens in new window" lang string
as its alt text and add it to the "Give feedback about this software"
link.
* Update cta template to indicate whether the link opens in new window.
* The basic_test::assertTag method will issue a warning as $tag is not
found, failing the PHP Unit test that uses this method.
* Add tests to check that assertTag is working
The joubel/core and joubel/editor libraries have been moved to Moodle
namespace.
This commit adds the new namespace to the places where these classes
are used.
Besides, a couple of minor changes have been done to replace the _test.php
classname and remove some unnecessary defined('MOODLE_INTERNAL').
The joubel/editor is a third-party library. A namespace has been added to
avoid collision with other plugins using it (such as mod_hvp).
That way, they will be able to have a different version without
side effects.
The joubel/core is a third-party library. A namespace has been added to
avoid collision with other plugins using it (such as mod_hvp).
That way, they will be able to have a different version without
side effects.
This value was being passed in the template, but the template
had 0 hardcoded instead of injecting the context value. With the other
bug fixes in this issue correctly loading the "All" view properly on
page load, this uncovered that at load time, the course view would not
load overdue items, which also meant a behat test was failing.
The 6 month option was highlighted on load if previously selected,
which was not the case for any other options. This has been removed
so it behaves consistently.
When the timeframe filter was set to "all" when the timeline
block was initially loaded, an incorrect value rendered into
the template meant the timeframe limit was set to 0 (which will
return no results) instead of setting no limit (which would fetch all
action events, as intended).
When using a try/catch for a find if the first match is not met then we
have an implicit 6 second delay waiting for the search to time out.
We can avoid this by combining the searches in a comma-separated
selection.
I have also taken the opportunity to convert the use of ->click() to the
i_click_on function which has the benefit of supporting wait for
pendingJS.
The standard NodeElement functions for getAttribute, getTagName,
getParent, and friends go back to WebDriver and parse the DOM for each
request. This is insanely slow per request, and in the case of forms we
do a lot of checking to determine the field type.
This change modifies the form field detection to copy the entire node
content into a DOMDocument and parse the document locally.
This is significantly faster - in some cases where there are large
documents minutes faster.
I believe that this should be a safe change as the document fetched from
the browser is normalised to match the doctype specified.
This was broken by changes in fdd5aef0, as the sesskey URL parameter
was required on admin/index.php in order for block related actions
to be performed uninterrupted by any redirections.
MDL-71899 produces a regression. When an activity plugin is not present
but it has activities in the DB the course view returns an exception.
With this patch the code will ignore missing modules when calculating
the completion.
Corresponding move of language string invalidpersistenterror following
move of class invalid_persistent_exception in MDL-57273.
AMOS BEGIN
CPY [invalidpersistenterror,core_competencies],[invalidpersistenterror,core_error]
AMOS END
With the default 5 sections course, after editing the activities, the
"log out" link is exactly on the edge of the view port. It's one of
those cases where Chrome fails to click/press it (have tried both).
So just make the course to have 3 sections is enough to avoid the
problem. Some day Chrome will fix that problem but, until then...
we cannot do much apart from changing the window size or try to
configure things to show differently.
All the scenarios failing with Chrome zero-size errors are
getting the @skip_chrome_zerosize to be able to keep them apart from
regular runs. See MDL-71108 for more information about
that error.
They will be run by another job, specifically using that tag to
keep them failing and under control.
In order to achieve good colour contrast between dimmed text and striped
table rows, the background colour of the striped rows are being
lightened to 0.03 alpha. While $gray-600 is being adjusted to a bit
darker shade of #6a737b. This gives a colour contrast ratio of 4.5:1
between the dimmed text colour and the striped table row background
which meets WCAG 2.1 Level AA requirements.
Comparing the normal font colour (#212529) with the adjusted dimmed text
colour gives a contrast ratio of 3.20:1 which meets WCAG 2.1 Level AA
requirements and still makes them distinguishable from one another.
- Modified SQL in get_data() to get the activity completion of all course modules without calling get_fast_modinfo() to reduce the DB queries.
- Building activity completion cache for the whole course in one time instead of building the cache for each course module by passing $wholecourse = true to get_data()
Fetching the counts of unread posts should only include unread regular
posts or unread private reply posts directed to the user unless the user
has the capability to read private replies.
In order to retrieve the correct counts, we also need to loop through
each forum instance in the course to check the capability of the user to
read private replies in each forum.
We should ensure that users being exported are enrolled on the
course being exported from. In courses where the current user can't
access all groups we should ensure that users being exported belong
to the same groups as them.
LaTeX documents have a preamble section and admins can use a \newcommand
statement to define new commands there (or to give an alias to another
command). This commit makes sure no blocked command can escape
sanitization by being used in a new seemingly harmless command that is
defined in the LaTeX preamble.
Before PHP 8.0.10, the time is lost when datetime->modify() is used
with weekdays. However, with PHP 8.0.10, the time is kept.
In consequence, the test_yearly_every_20th_monday_forever method
started failing.
For fixing it, the offset has been removed and the time has been set
to 00:00, so the behaviour will be the same regardless the PHP
version.
When npm-shrinkwrap.json was regenerated few days ago by MDL-72014
running audit-fix on it, it seems that some checksums in the "lodash"
package were missing.
Now they are properly set and that's leading to npm-shrinkwrap.json
changes when people runs npm install (note npm ci is the recommended
way to install all the dev dependencies).
So, just reconcile the file to include the new checksums.
In MoodleCloud it was raised that, in some cases, loading the content
bank, from a course page, gives an "Invalid H5P content URL", not
offering any way to delete the offending content or create new one.
An extra-check has been added to the "is_view_allowed" method to
guarantee the H5P API is called only if the H5P content has a file.
This optional environmental check will look if:
- the site is running php72.
- the site has igbinary extension enabled.
- the igbinary extension version is a buggy one >=3.2.2 <= 3.2.4
- the bug is reproducible.
And will warn if all the conditions are met.
When you click on 'Edit' then 'Hide' on a grade item like a course category
that contains many grade items, the course and the grade items are set to
hidden. However, if you use the 'hidden' option on the 'Edit settings'
page it only hides that item and none of the items (assignment etc)
under that course category. We want them to be consistent.
When you click on 'Edit' then 'Hide' on a grade item like an
assignment it also hides the grades. However, If you use the
'hidden' option on the 'Edit settings' page it just hides that
item and doesn't hide grades as well. We want them to be
consistent.
The course unit test assumes the number of db call. This assumption
is correct in vanilla moodle only. Once some plugins installed their
observers may do extra db calls. Due to caching the number of actual
db calls in the unit test may be fewer.
Fetching user tours used to fail on external_api::validate_context() and
require_login() calls if the user did not have the site policy agreed.
The patch introduces a check to see if the user is fully set up and
ready to use the site before attempting to load the tours.
moodle_read_slave_trait: when creating another handle, restore temptables
property that is clobbered by raw_connect().
Also a better condition for temptable related queries detection in
pgsql_native_moodle_database.
dml_pgsql_read_slave_test::test_temp_table(): use real db connection
if possible, otherwise skip the test.
It was a mistake to force filtering of SVG files in MDL-55243. It can
easily lead to corrupted SVG files.
The patch removes that forced filtering and clarifies the inline comment
of what and why we need to do.
Where there are no grade items with idnumbers for selection, then
nothing should be exported. Currently, this is being interpreted as
the user exporting all grade items.
Before, we had each redirect test duplicated: one for the native
redirects via native cURL, second for our emulated implementation. Now
all redirects are always emulated so there is no need to have them
tested twice.
The security problem here was that only the first and the last URL in
the redirect chain was checked by the security helper. This patch forces
the curl wrapper to always emulate cURL redirects and check every
redirect URL in the chain before actually visiting it.
The new parameter of curl_security_helper::url_is_blocked() introduced
in MDL-71916 became part of the API. Even if we reverted it quickly,
someone can use a released Moodle version that has that parameter in
place. For that reason and also to avoid potential troubles in the
future (e.g. when yet another argument would be added to this method),
we need to make it clear that the second parameter of this method should
never be used again.
Poor $maxredirects, you did not live long with us. Oh well.
This reverts the original fix introduced in MDL-71916. It introduced an
extra native cURL call inside curl_security_helper to check if the given
URL triggers a redirect to a blocked URL or not.
Shortly after the release, a couple of regressions were reported as a
result of the integrated solution. It was agreed to revert the fix and
progress with implementing an alternative approach.
The simple pattern matches were conflicting in some situations. To make
this backwards compatable we need to convert it to a Regex pattern match
instead, and provide the quoted and unquoted variants.
There were two issues here:
* I am on the [categoryname] category page page (duplicated page)
* the wrong URL was being used
Since an incorrect URL was used I felt it safe to rename the step from
'category page page' to 'category page'.
This commit makes the following improvements to core page resolverss:
* allows for mixed case naming (course, Course, etc.)
* allows fields other than the idnumber to be specified:
** course: idnumber, shortname, fullname
** course category: idnumber, name
Whilst some of these fields are not unique, they will typically be
unique in most test scenarios. Where they are not then the idnumber
should be used in preference.
Move the close button out of the element that is set as the
aria-labelledby for the dialog.
Also used h5 for the dialog titles so they are consistent with
AMD modals.
The initial focused element should be the first operator button.
This commit also fixes the accessibility issues that previously were
wrongly fixed by a redundant click on the first tab using javascript.
Clicking in a day in the three month block replaces the div with id starting
with calendar- so we can do so repeatedly and successfully and test that it does (with behat).
The activity generator currently requires an idnumber when creating
activities, but this is not a requirement when creating the same
activity through the UI. The requirement comes because we want to
provide a way to refer to activities in subsequent steps.
This commit modifies the behaviour such that the generator uses the name
of the activity as the default idnumber.
This has two main benefits:
1. it simplfies generation of activities; and
2. it makes the language used when writing behat tests much more natural.
With this change, steps will refer to the activity by its idnumber/title
in all cases, rather than sometimes by an idnumber which bears no
relevance to the title.
The `page_type_list` method for the message component was deprecated
in 5b0769db as part of MDL-54744. However it is still required when
trying to configure blocks on any messaging pages.
After a major upgrade was done in h5p.com, some random errors appeared
in the "H5P options are ignored for H5P URLs" scenario.
They have been fixed replacing the URL for different (which should
load quicker). As we're checking external content, no other improvement
can be done on the Moodle site.
Take account of parent languages when requesting given library
translation. For example if we are currently using "de_kids" as the
current language, we need to recurse each language pack looking for
a matching H5P translation ("de_kids" -> "de_du" -> "de").
A similar approach was followed @ MDL-67935, although the
case doesn't seem to be the same.
In any case, it's curious that all local execution are
100% passing and only CIs fail so often. Just guessing if,
maybe, there is some Chrome version factor around (we
are running older, sticky, versions @ CIs.
Adds new unit test, test_course_get_recent_courses_sort_validation(),
which is reponsible for testing the validation of the sort value in
course_get_recent_courses().
When a teacher changes the subscription mode from forced
subscription to auto subscription using the "cog" icon in the
forum page, users are not seen in the Show/edit current
subscribers" page.
This does not happens when done from the Edit settings page.
Add option to class csv_export_writer to prefix CSV file with UTF-8 byte
order mark (BOM). This helps Microsoft Excel detect the file's character
encoding.
This patch fix the subcription page only adding the course parameter
only when viewing the calendar on course context. It also passes the
url to the form instance, so we have the course parameter on the
form action attribute.
The table button can become a menu button when the cursor's position
in the editor is within a table. So we'd need to update the button with
ARIA attributes appropriate for a menu button.
This is best done when the button's highlight gets toggled, so we're
adding an event listener for when this happens and add/remove the ARIA
attributes accordingly.
* Menu button fix
- Added aria-haspopup, aria-controls, and aria-expanded attributes.
* Menu fixes
- Added aria-labelledby that points to the menu button label.
- Removed the dialog role in the menu's container.
- The ul tag needs the "menu" role.
- The li tag needs the role "none" instead of the "presentation" role
Reference:
https://www.w3.org/TR/wai-aria-practices-1.1/#menubutton
Do not allow maxgrade change when some of the
users are already graded. As of now this is applicable
to the following activities:
1. Forum
2. Database
3. Lesson
4. Glossary
Signed-off-by: Sujith Haridasan <sujith@moodle.com>
This change set would bring the following new additions
to the nextcloud repo:
1. Create a new radio button in filepicker: "Link to file"
2. When user clicks this radio button a warning message
would be created, saying this file would become public.
Meaning a public link is created in the nextcloud server.
3. Created a sync_reference method to sync the files downloaded
from nextcloud server. The sync/refresh time given is 1 day/24 hours.
4. Made sure that when the file is downloaded, we use the file
from moodledata file pool.
Signed-off-by: Sujith Haridasan <sujith@moodle.com>
The deprecated method to render the dropdown based activity chooser
from 430746d3 was still being called, which produced debugging on
the site when doing so.
It's mostly addDocuments(), used by test_add_document_batch_large()
with 100 big documents what requires a lot of ram
although apparently, it's freed (partially) once ended.
Just the peak usage remains high. In any case, isolating that
test to avoid the non-freed side of it to consume too
much memory for the rest. (We allow "only" 2GB).
Still, I think that there is a good work about to detect which
tests are leaking / consuming too much, I'd bet there are
a bunch running completely out of control.
Upstream behat/mink-extension isn't updated since 2018.
So we switch to friends-of-behat/mink-extension until
upstream situation is clarified.
But friends-of-behat/mink-extension has some PHP8 issues
that, at the time of writting this, have not been solved:
- https://github.com/FriendsOfBehat/MinkExtension/pull/9
So we are temporarily changing to one fork:
https://github.com/moodlehq/MinkExtension
Whenever those issues are solved by friends-of-beha or behat
we will change back to them and remove our fork.
Generated with composer 2.x and PHP 7.3
In certain operative systems the browser's scrollbar may partly cover
the content in the student header and user cells when horizontally
scrolling through the table contents. This is most noticeable when in
RTL mode. Adding slight padding on the left (or right in RTL mode) to
the content in these cells would fix this UI issue.
With significant-digits presentation mode, NAN would be formatted as
'NAN' followed by zeroes, which looks silly, and +INF would cause an
execution timeout as Moodle tries to divide +∞ by 10 until it becomes
less than 1, which can happen in a divide-by-zero situation.
Note that the user can't answer NAN or INF to any question, but at this
at least now looks consistent and also doesn't break Moodle in the case
of +INF.
The previous default value for this config was based on the
PHP ini file `display_error` property (6349a3ba). However we
override this propertly during page setup (25338300) according
to the value of the configuration itself.
This had the effect of always setting the default value for
this config to it's current value.
Issue 1: While essay question's uploading progress, we need to disable submit
buttons to prevent submit form event.
Issue 2: Enable buttons after pressing cancel button on the popup
confirming overwrite file existed.
The step "And I wait until ".block_myoverview
[data-control='next']" "css_element" exists" is not correct
because this [data-control='next'] element exists before and
after the step so, in some cases, it might cause the following
step will start earlier than expected.
As pending JS has been added, this wait steps are not required
any more.
Activity modules may not have the associated grade_item created yet. It
used to throw fatal error in that case - even when trying to view the
course or edit the activity. So there was no easy way to recover from
this situation.
The patch is based on reasoning that an activity without grade item is
same as activity without any grades. And as such it is considered
incomplete.
A new unit test is added to cover this specific scenario. The existing
unit test is modified and it does not expect the exception any more.
There does not seem to be any good reason why this situation should be
exceptional.
Some fields were being hidden based on whether the service was set up
for internal service use or was a service shown on the login page.
These fields refer to OAuth 2 logins and must be configurable for both
kinds of services.
It seems that the new phpcs3 checker is now controlling those
line comments that previously were ignored.
This commit just looks for all the cases and bulk-add
them when needed. The bash script (mac) used to add all them is:
while read -r line; do
arr=(${line//:/ })
if [[ -n ${arr[0]} ]] && [[ -n ${arr[1]} ]]; then
echo " file ${arr[0]}, line ${arr[1]}"
sed -i "${arr[1]}s/\$/\./" ${arr[0]}
fi
done < <(find . -name version.php | xargs ag --nomultiline '>(version|requires) *=.*//.*[^;\.]$')
version = 2021051700 release version
requires= 2021051100 current rc1 (week7roll1) version
Note that, because we are under parallel development period,
this is being done in the branch that is going to be released
(MOODLE_311_STABLE already existing) for Moodle 3.11.0 and
not in master, that is the one getting the bump under normal
(non-parallel) periods.
Also adds xmlrpc-beta from PECL (PHP8 moved the extension there)
and setups max_input_vars=5000 as initial value (see MDL-71390).
Note that this also includes a tiny modification when configuring
redis sessions, because there was some outdated code from old issue,
now fixed (MDL-60978), so that info has been removed.
Without this, people can craft URLs that other users might use not realising
what they do - and as a XSS vulnerability, it could do any number of things the
clicking-user has access to do on the site.
Change-Id: I82adc71e8706d8929011b4b24523d5b62b8ccea1
* The button to "Add group/grouping access restriction" under
common module settings should only be present if corresponding
availability plugins are enabled; and
* Prevent the same button behaving as a submit button, which
intercepted the user hitting return in the form and added a
restriction without deliberate action taken by the user.
This isn't really a deprecations where something is replaced
by another and all uses must adapt to. Instead it's just a
fallback functionality that will stop working in Moodle 4.2.
Because of that, I've tried to add TODO/@todo comments to
the places that will need to modified, always pointing to
MDL-71421 that is where the removal will happen.
Also, haven't added any debugging() output, after thinking
a lot about it, because this isn't anything that developers
can be using but a internal implementation detail (a fallback)
that we want to remove in some versions.
1) The GROUP BY statement must include all fields in the SELECT
not being aggregated or constant cols.
2) There was a missing space causing problems to databases
using named parameters.
In order to allow for correct seb:// or sebs:// calls without browser
warnings of insecure links, it is not possible to send a get request
with an attached cmid parameter to the unknown seb:// or sebs://
URL via a form button.
We've got to use a <a href> link outside a form to circumvent
browsers warning of an insecure link and call Safe Exam Browser
correctly.
Table names and database name now enclosed in backticks.
admin/cli/mysql_collation.php failed if $CFG->prefix was blank with
MySQL 8.0 because table 'groups' conflicted with a new reserved word.
Note that this quotes both mysql_collation.php and mysql_compressed_rows.php
but only the former was mal-functioning.
The case of mysql_compressed_rows.php is a little bit special because
not all tables are processed, only those having big rows. And the groups
table is not one of them. In any case, better add the quotes there for
any future case.
Also, when testing this patch https://tracker.moodle.org/browse/MDL-71512
was discovered and will need to be fixed to make core 100% compliant
with MySQL 8.0 and MariaDB 10.6 and up.
Previously missing were the two new properties added in 36bc7b07:
showactivitydates & showcompletionconditions - this change ensures
all properties are requested before exporting courses linked to
competencies.
- The mbstring extension is required since Moodle 3.9 onwards so
the upgrading steps have been removed because they are not
required any more.
- The pull-request for fixing some minor PHP7.4 problems has been
integrated so this patch is not required.
Other tests check behaviour of individual conditions, this tests that
when multiple core and custom conditions are required, each will be
updated as required and not cache until all are completed.
When an automatic completion condition may have had its state change,
we now unset the cached value for the user's completion in the relevant
activity, so up-to-date values are re-fetched and available to students.
The previous behaviour was that custom conditions would remain cached
until the activity reached overall completion.
File and folder names cannot end with dots on Windows. So replace the
trailing dots with underscore consistently with how some other zip tools
(such as 7-zip) handle this case.
Creates a new capability mod/assign:viewownsubmissionsummary which will
be required in order to display the submission summary to the user in
the assignment page.
Allow for conditions that are not set, consistent with other modules.
They will usually always be set in practice, but best to handle it,
which also helps with unit testing.
This adds the requirement for activities supporting custom completion to
specify the order all completion conditions should be displayed for that
activity. It also implements the sorting that takes place.
The quiz module now showa the quiz open and close dates at the top of
the page. So there is no need for the quizaccess_openclosedate plugin to
display these dates in the middle of the page.
The $extradetails parameter in the constructor of the
mod_feedback\output\summary class is no longer used by any of the codes
that instantiate the class. Therefore, it is removed.
The respective template is updated as well.
AMOS BEGIN
CPY [relativedatessubmissionduedateafter,mod_assign],[relativedatessubmissionduedateafter,core_course]
CPY [relativedatessubmissionduedatebefore,mod_assign],[relativedatessubmissionduedatebefore,core_course]
AMOS END
The define_after_data method was looking for a required_param
when it should have been an optional parameter. This was causing
error messages when creating a new datetime field.
Provide page for the reports in the course administration.
The page has a drop down to navigate to different reports.
The navigation is remembered just like the grade reports in
the course.
Signed-off-by: Sujith Haridasan <sujith@moodle.com>
Hidden courses can be used for training
but we do not want to generate insights for them
because students do not have access to hidden courses.
This was fixed in MDL-66806 for "Students at risk" model.
Fixed for "Students who have not accessed the course recently" in this issue.
1. pg_connection_status(false) fails on PHP8 because the argument is not a stream
2. skipeed the unittest that expected that pg_query() does not show an error if the resource is invalid
see also MDL-71482
Instead of using the file title when generating the source attribute,
we should be using only the name of the Google Drive file in order to
avoid having the original extension within the filename after the file
is imported and added into Moodle.
Current test works ok every day of the year but the day of
DST begin changes (last Sunday of March currently).
That day, 01:00 Europe/London simply doesn't exist, because
the time is moved forward by 60 minutes.
Hence the expectation for the next cron execution is not 1 "normal"
day (24h) but 1 less hour.
Note this only happens for the current test on DST begins, the
opposite change (DST ends), happening the last Sunday of October
is not affected, because, at all effects, 01:00 is a perfectly normal
and existing hour.
We update the dates with user/group overrides. The calculation of
the override in the quiz module is different from the assignment
module as the quiz_overrides table des not have a sortorder column.
See quiz_update_effective_access().
We update the dates with user/group overrides. The calculation of
the override in the lesson module is different from the assignment
module as the lesson_overrides table des not have a sortorder column.
See lesson::update_effective_access().
Replicates changes made to notification preferences in e41b3485.
The templates for each are similar enough than one can be used
from the other. Remove redundant styles for previous controls.
* When completion tracking is not enabled for the course, it does not
make sense for the course's showcompletionconditions setting to
be set according to the default value indicated by the
"moodlecourse | showcompletionconditions" admin setting. Setting
showcompletionconditions as enabled when completion tracking is disabled
makes even less sense. So in such a case, we should not be setting a
default value for showcompletionconditions and allow it to be null.
* When the course is edited and completion tracking is enabled, this
also would set the "Show completion conditions" field to default to the
value set in the "moodlecourse | showcompletionconditions" admin
setting.
* *_get_completion_state() callback functions have been deprecated and
should no longer be used. Plugins that define custom completion rules
must implement the mod_[modname]\completion\custom_completion class that
extends the \core_completion\activity_custom_completion base class.
The machinery to fix orphaned calendar events that were broken by MDL-67494.
The solution consists of:
1) Upgrade step that checks if this site has executed the problematic upgrade steps and
if positive, it will schedule a new run for calendar_fix_orphaned_events adhoc task.
2) Adhoc task that will self-spawn calling the recovery machinery, running until
all the orphaned calendar events are fixed. It also sets the maximum runtime of
60 seconds. It is also possible to override that number by specifing the desired
number setting the ->calendareventsmaxseconds in your config.php
3) CLI script that will look for all the calendar events which userids
where broken by a wrong upgrade step, affecting to Moodle 3.9.5
and up.
It performs checks to both:
a) Detect if the site was affected (ran the wrong upgrade step).
b) Look for orphaned calendar events, categorising them as:
- standard: site / category / course / group / user events
- subscription: events created via subscriptions.
- action: normal action events, created to show common important dates.
- override: user and group override events, particular, that some activities support.
- custom: other events, not being any of the above, common or particular.
By specifying it (--fix) try to recover as many broken events (missing userid) as
possible. Standard, subscription, action, override events in core are fully supported but
override or custom events should be fixed by each plugin as far as there isn't any standard
API (plugin-wise) to launch a rebuild of the calendar events.
4) Unit tests and helper functions to generate calendar events. We have decided to
keep the tests simple, testing only true and false and not using data generators because
the code is purely to recover the calendar events and won't turn into an API or something
and also due to the urgency of this issue.
The helpers have been created in calendar/tests/helpers.php since there are no data generators
for calendar.
This commits adds a fallback for plugins which does not have
custom_completion implementation.
For those cases, it will search for {modulename}_get_completion_state
callback in the plugin and call get_overall_completion() method in
cm_completion_details class to get the overall completion state for
a course module and user.
The preferences table did not meet accessibility guidelines regarding
colour contrast between hovered rows and the disabled switch
elements.
Ensure white background is used for all table cells, and replace the
highlighted columns used to differentiate processors with borders.
Before this change a teacher would be able to see users listed if:
* They have an active enrolment and can submit
* They have an an inactive enrolment for a role that can submit
After this change they will additonally be able to see users listed:
* That have an active enrolment and have submitted
* That have an inactive enrolment and have submitted
This means that if an assignment has it's context frozen all users
that have made some form of submission will still be listed.
It will also apply if the submission capability is removed from a
user.
If a user's enrolment is deleted they will not be listed.
The submission and grading counts have also been updated so
they will reflect the new rules.
Before this change if a student visited an assignment that is
frozen they would only see the title and description even if
they had made a submission to it.
After the change they will be able to see the status of their
submission and any feedback and grades they have recived.
It will also make the Moodle app recognise that submission
should not happen because the assignment is frozen.
Tests based on ones created by Andrew Nicols
- some engine error messages changed
- some warning levels changed
- the carriage return symbol randomly appears
- one phpunit assertion fails and not really representative of anything
Create a new profile field type, move all existing content of the fields
'icq', 'skype', 'aim', 'yahoo', 'msn' and 'url' in the mdl_user table to
theses new profile fields if needed.
AMOS BEGIN
MOV [aimid,core],[aimid,courseimage,profilefield_social]
MOV [yahooid,core],[yahooid,profilefield_social]
MOV [skypeid,core],[skypeid,profilefield_social]
MOV [icqnumber,core],[icqnumber,profilefield_social]
MOV [msnid,core],[msnid,profilefield_social]
MOV [webpage,core],[webpage,profilefield_social]
AMOS END
In PHP8.0 using `ksort` was producing incorrect results by sorting
keys differing only in case in the wrong order. This change makes
sorting consistent between PHP versions.
Co-Authored-By: Tim Hunt <T.J.Hunt@open.ac.uk>
Backup the table user_lastaccess which holds data for when a user last accessed a course.
And also restore to the table user_lastaccess from the .xml file created in the backup.
As part of the tests preparations, some backups are created, but
the tests aren't calling to their ->destroy() method at the end.
The destroy() method is needed to go cleaning and unsetting various
objects that, only then, will free resources, like memory, open files...
This isn't critical under Unix OSs... but Windows is really sensible
and keeps all open files locked, so any write operation on them fails.
- Some small namespace/use/MOODLE_INTERNAL/requires reorganization.
- Move file phpdocs to class phpdocs.
- Move external method names to be new "execute" default ones.
- Move unit tests to matching core_grades\external namespace.
- Add missing clean_returnvalue() calls.
This adds a new webservice that creates gradecategories as a batch,
and deprecates the old single creation call, which is superseded.
It also patches a few small issues in the single WS, for any integrations
currently relying on that webservice.
Some recent tests do set a date time element
to ##now## or tomorrow and, immediately after that
the look if, effectively, ##now## and #tomorrow#
have been set (with minutes resolution).
Problem is that, between the field is set and the field
is verified, it can happen that the time advances to
next minute (from H:M:59 to H:M+1:00) and then the
assertion fails.
To avoid this, we could have lowered the resolution to be
hours... but that doesn't solve the problem just makes it
to happen less often.
So, instead of that... we are setting the 2 now and tomorrow
cases to be "today noon" and "tomorrow noon" (12:00:00) so
we ensure they won't be ever in the risk of jumping of minute.
* The activity information output component is still being displayed
even in the case when the activity does not have dates to show and
completion tracking is not enabled.
This also causes a horizontal line to be rendered in activity view pages
below the activity heading which does not serve much purpose when
there's no activity information to be shown.
The activity information output component should only be displayed
if there's completion information to show or if there are activty dates
to show. Otherwise, there's no need to render it.
Even if a prediction is hidden from the report once is flagged,
it can be flagged several times if the user visits detailed view via URL.
We remove the checkbox to select a prediction and flag it
once it has already been flagged.
Some of the pre-existing behat tests were creating content into the
content bank using the generator but the H5P content-type libraries
weren't uploaded.
Since MDL-69331, libraries need to be deployed previously (so fake
H5P content-type won't be displayed if these libraries don't exist).
If a H5P content-type is disabled:
- The content bank won't display existing contents having it as a
main library.
- The content bank won't allow to create new contents using it.
The addition of the activity information moves the h5p content down
out of the window view on Firefox which causes the Check button to be
unclickable. This patch increases the window size to ensure that the
check button is visible and clickable.
Credits to Eloy <stronk7@moodle.org> for the thorough investigation
of the root cause of this issue :)
The form to upload/replace files has been moved to classes/form
(it's where it should have been added when it was created).
Besides, this class has been renamed because now, with the
namespace, contentbank and form don't need to be in the name ;-)
Augment all installed plugin's security helper if there is any.
Plugins function has to be defined as plugintype_pluginname_security_helper in pluginname/lib.php file.
Some errors raised when calling external services were ignored.
Displaying this information helps to debug and find the real reasons
why some action, like connecting to an external backpack, can't be
done.
The scopes parameter should be passed when creating the OAuth2 badges
client. As it is an optional parameter, when it's empty, it will
be initilised with the supported scopes for the backpack issuer.
It will happen, for instance, when a call is done to oauth2callback.php
because, as defined in RFC6749, OAuth2 authorization response only
supports code and state.
Render the activity information output component in the course homepage
only if either completion details or activity dates are to be displayed.
This can help reduce the number of files being included when loading the
course homepage (e.g. the activity information template for each
activity in the course homepage).
Ref: https://php.watch/versions/8.0/final-private-function#final-private
This applies #712 from upstream libraries.
Unrelated: I've also changed a occurrence of "private final"
by the correct (PSR-12) "final private" in the constructor
(the final&private combination is allowed in constructors)
in lib/classes/event/base.php, because it's the unique case
in core and phpcs was really insisting.
* Make sure the activity is visible to the user (cm_info::uservisible)
before showing the activity completion information.
* Add to-do status for overridden automatic completion
* Check the activity dates on the course homepage depending on
the value of the showactivitydates course setting
* Plus use the new Behat steps for checking activity dates
* activity_date_in_activity_should_contain_text()
- Checks the presence of the given text in the activity's date info.
* activity_dates_information_in_activity_should_exist()
- Checks the presence of activity dates information in the activity
information output component.
* activity_dates_information_in_activity_should_not_exist()
- Checks the absence of activity dates information in the activity
information output component.
* activity_completion_condition_displayed_as()
- Shows the status of an automatic completion condition for a given
activity
* activity_should_have_the_completion_condition()
- Checks that the activity displays the given completion condition
* manual_completion_button_displayed_as()
- Shows the status of the manual completion button for a given
activity
* the_manual_completion_button_for_activity_should_be_disabled()
- Checks that the manual completion button for the given activity is
disabled.
* there_should_be_no_completion_for_activity()
- Checks that the activity does not show completion information
* toggle_the_manual_completion_state()
- Toggles the manual completion button's state for a given activity
* overridden_manual_completion_button_displayed_as()
- Checks the manual completion state of an activity.
* overridden_activity_completion_condition_displayed_as()
- Checks if the activity with specified name is maked as complete.
* Add activity name for completion conditions labels. This would give
better information to screen reader users the activity that the list
of automatic completion conditions belong to. This would be useful
especially when the completion conditions are displayed on the course
homepage.
* Add data-region attributes to activity dates and completion
information divs.
* Reorganise activity dates and completion information divs so they
are only rendered when they have data to show.
Deprecate \core_course_renderer::course_section_cm_completion(). It is
not being used anymore and is being replaced by
\core_renderer::activity_information().
When an activity has manual completion tracking, pressing the manual
completion checkbox reloads the page after toggling the completion
state when the activity is linked to availability conditions.
The "Mark as done" button needs to mimic this behaviour as well.
The approach being taken here is to add a core_course/view JS module
for the course homepage which listens for the manualCompletionToggled
event and reloads the page when the activity module has availability
conditions tied to it.
Perhaps for future development, instead of reloading the page, the
container of the restricted course sections/activities can reloaded via
AJAX as well.
That way we can check that the expectations are
working ok for every database. Previously only the
ANY case was being tested.
Also, convert them to course badges and check, via
review_all_criteria() that the SQL structures returned
by award_criteria_cohort->get_completed_criteria_sql()
doesn't fail and return the expected counters.
When buildilng the 'WHERE' clause, the SQL generated
a query that selected any user who was a member of the
**last** group, not all of the groups.
I believe the query has to be re-worked so that it only
returns users who are in ALL of the groups. This can
be done by a GROUP BY and HAVING.
Try to correct broken ul/ol/li tags, as they have an outsized impact
on course layout. Uses basic regex and loops to track open and closed
tags.
Also adds a deep clean option to the HTML cleaner, that runs less
frequent, more intensive cleanings. This is because normally _cleanHTML
gets called after each keystroke, which could cause problems with
large content on weak systems.
Behat changes are a fix for setting multiline strings in Atto, and
creating a multiline match step.
'usercreated' and 'usermodified' fields can not be always mapped.
We are filling those fields with 'old' users when working on the same
site, or with current user when working on a different instance.
Custom rules are combined using the AND operator. However, the
'completionpass' and 'completionattemptsexhausted' rules are exceptions
because they are combined together with OR, and form a single new rule
that then will be combined with the rest of the rules using AND.
This commit combines the mentioned rules into a single rule named
'completionpassorattemptsexhausted'.
This includes 2 change to the order date(time) elements are filled,
each one addressing one type of problem, where current order is
problematic and can lead to unexpected dates.
1) Changing date, when current month only has X days and target
month has more than X days. Example, being 1 April, change
the date to 31 May.
This is solved by changing the order of introduction
from current D => M => Y to Y => M => D.
2) Changing date, when target month only has X days and current
month has more that X days. Example, being 31 March, change
the date to 28 Feb.
This is solved by always setting the D to 1, before the
Y => M => D sequence commented @ 1) begins.
Because of the order that dates and datetimes are filled by behat
sometimes there are some intermediate results that are impossible
and then the form (javascript) automatically reacts and fixes the
date, ultimate leading to a different date that the one we wanted
to set with Behat.
This is noticeable when switching between months (with some day
being the last day of the month) and the 2 months have a different
number of days. For example April date => March date (or the opposite).
This test covers all the critical changes (day, month and year),
back and forth. All times are Perth/Australia.
By rounding the current time it was possible that the most recently
created user enrolments (e.g. self enrolments) were being excluded.
This would manifest itself in a user being enrolled on a course,
but it not appearing under "My courses" navigation or on their own
Dashboard until the rounded time had caught up with the current
time.
The `isset` call always returned true for the 'replace' option,
because even if not specified it receives a non-null value. The
`strpos` call now does a strict equality check rather than
greater-than-or-equal (which always returns true).
This change removes references in code and comments to a few
deprecated functions that were accidentally missed in the larger
change.
The code changes only affect service_users.php. I made it support
custom user profile fields in this query, because it was easy
enough, rather than adding another TODO to fix later.
The activity information output component displays information about
an activity module that can contain:
1. Activity dates
2. Completion information
a. A manual completion button; or
b. A list of automatic completion conditions and their statuses.
This patch also includes a new JS module called
core_course/manual_completion_toggle for toggling the
completion state of activities that support manual completion.
Allow the core_completion_update_activity_completion_status_manually
web service function to be called from AJAX as well. This is needed
by the new manual completion toggle button in the
core_course/completion_manual template.
Due to Solr bug SOLR-15039, uploading files for indexing can fail if
it uses multipart upload. This changes it to use direct binary upload.
Unfortunately, the direct binary version in PHP curl only supports a
string, so we have to load the file into memory. I added extra code to
restrict the size of files indexed to (memory limit - 100MB), which is
usually 284MB unless configured differently because cron runs under
MEMORY_EXTRA.
Before resetting completion states during module update, we need to
rebuild the course cache first in order to properly reset the completion
states. Otherwise, calls to methods that fetch course module info
via cache (e.g. cm_info::create()) will fetch outdated information.
Note that there wasn't any case of multiple testcase classes in
1 file. All the cases reported in the issue were false positives
caused but other "mock/fixture" files being named _test.
So all this issue does is:
1) rename any _test suffixed class in test files, because we are
going to start renaming a lot of test classes to _test.
2) ensure that the 2 test case classes modified in this issue,
are already observing the filename = classname rule that will
be implemented soon (and verigy it works).
This patch adds some changes to guarantee that OpenBadges 2.1 (OBv2.1)
will return always the same JSON format that OBv2.0 (because the only
difference between both versions it's the BadgeConnect API).
By not redirecting when we are already on the purge caches page we
can preserve the submitted form data, allowing users to easily
repeat the previous action.
In MDL-70689, Eloy and Helen suggested some fixes to improve this patch.
As this is a followup issue to move the pending services, this
very first commit will fix pending things raised in the parent issue.
They are thrown in the following circumstances:
* The dynamic form class doesn't exist;
* It does exist but it's `check_access` method throws exception
Co-Authored-By: Andrew Nicols <andrew@nicols.co.uk>
This class would belong more appropriately within the 'user' API
(core_user) instead of within the 'core' API, since it is
directly related to user data.
Since the class has only just been added to Moodle, now is a good
time to move it.
This commit adds the missing logic for category events in
get_context() method. Without this fix, users will get errors
if they have category events with userid set to 0.
We should be able to fetch the grade completion status for modules
that do not have custom completion configured. This also improves
unit testing coverage.
* Fixed inverted $sameuser test data.
* Fixed caching expectation check. Caching only relies on whether the
user accessing the completion data is the same user or not.
* Fixed checking for the caching of other modules. Should have been
checking cm ID and not instance ID.
* Additional test case when whole course parameter is passed as
true, but the requesting user is different from the target user.
Core and mod_hvp are using the same namespace for some H5P
libraries. This is causing some random errors.
In order to get the expected behaviour in Moodle core, this patch
prepends H5P libraries in order to guarantee they are loaded first.
Plugins using same libraries will need to use a different namespace
if they overlap and a different version of these libraries should
be used.
The given condition in the if statement did not take into account cases
when the setting is locked by hierarchy. Settings which are locked by
hierarchy should should keep their current values which were inherited
from the parent.
Prior to this change the Grunt stylelint command was too greedy when
determining which files hsould be linted.
This change modifies the watch command to only watch relevant files and
subdirectories of each component directories. This means that unrelated
CSS and SCSS files are no longer watched for changes, and has the added
benefit of significantly increaseing the startup speed of grunt.
Without this patch applied the watch tasks were checking for matches in
the node_modules, and vendor directories.
Final deprecation for constant BADGE_BACKPACKURL (it has been removed)
and badges/backpack.js file (it has been removed too).
Besides, some methods have started the deprecation process too:
- badges_check_backpack_accessibility
- badges_setup_backpack_js()
- badges_local_backpack_js()
Apart from that, string 'addbackpack' has been deprecated here
because it's not used anywhere.
Backpack URLs (web and API) can have up to 255. However, the form
for creating/editing them, doesn't allow URLs with more than 50
characters. This patch align limitation to DB value (255).
Prior to this change all Grunt features were in a single Gruntfile.js
but this has become difficult to manage and maintain.
This commit moves the existing dependencies for component calculation
and babel moduel definition into a new .grunt directory, and
restructures the existing tasks in Gruntfile.js into separate task
configuration files.
This improves the maintainability of the Grunt build system and allows
for easier future expansion.
These recursive calls didn't work in PHP 7.3 and below, but in PHP
7.4 they also cause a fatal error which means if you have invalid
availability data, the whole page might die.
Upgrade mimetex to version 1.77. The steps followed
remained the same described in readme_moodle.txt.
This time we have made support for 64 bit architecture.
Hence all the binaries created now work with 64 bit only.
Following are the binaries which are upgraded:
1. Linux
2. Arm(aarch64)
3. Windows(exe)
4. Freebsd
5. Mac(darwin)
Signed-off-by: Sujith Haridasan <sujith@moodle.com>
Both in unit tests, but also in other areas (say blocks, for example)
it's possible that page->url has not been set yet.
So, let's change current code to always check if we have something
(page->url) to compare.
Note the use of has_set_url() is needed, because page->url is magic
method and it doesn't work as expected. Hate magics!
The link to the calendar page is missing in the navigation block in
classic which causes a behat failure in the 'Set availibility dates
for an assignment' feature. The fix includes adding the calendar block
to the Course page which can be used to navigate to the calendar page.
We used to display capabilities like "Manage any calendar entries",
"Delete evidence", "Manage competency frameworks", "View hidden courses"
and others as "Missing capabilities" for the mobile app service tokens.
This gave dangerous impression that the app will not work for students
without these capabilities granted. There are known cases of admins who
started to grant all these caps to the Authenticated user role because
they were afraid the app would not work for them.
The problem here is that the official mobile app service includes some
functions that have these capabilities declared as required. But they
are not really required to use the app. Either the app makes its own
clever checks of capabilities before calling the functions, or sometimes
the capabilities are not even correctly declared.
It is safer for everybody to display this information for custom
services only where the risk of the falsely missing caps is lower and
the information is more accurate.
Also, the help text has been improved so it does not suggest that these
capabilities must be always added. We do not know why the service has
them declared. In some cases, a service has capabilities declared just
because it makes use of them in the if-then fashion.
Additionally, the patch also displays the service short name because it
is actually needed to know.
The patch adds ability to filter the list of token by the token value,
the user and the service. Also the button to create a new token is made
more prominent and easier to spot.
No need to have a dedicated admin_setting_managewebservicetokens admin
setting subclass, and yet do not use almost any of its features. So the
patch merges the list and the forms handling into a single external
page.
Similarly, it feels like overkill to have a renderer method for a
simple confirmation widget - especially given that the URLs were
hard-coded in it. So that one was dropped.
The patch improves the display of the missing capabilities on the Manage
tokens page and the Select authorised users page. The list of
capabilities is rendered in a collapsible area and does not occupy too
much space by default.
Additionally on the later page, the patch improves the display of the
list of existing authorised users so that it respects the configured
identity fields.
The previous implementation falsely reported all implicit capabilities
inherited from the authenticated user archetype. That caused a lot of
capabilities reported as missing, even if they were correctly granted.
This new implementation uses a different logic. Instead of seeking for
explicitly assigned capabilities, it searches for capabilities that are
not assigned to any of the user's role across the system.
Please refer to the inline documentation. This should be still used for
informative reports only, not for actual permissions evaluation. The
context has been ignored here, as well as all the overrides etc. This
patch just makes it a lesser evil.
The purpose of this external function is to provide data for
asynchronous user selectors and similar widgets. It allows to search
users matching the given query in their name or other available identity
fields.
I suppose the original reasoning was that the web services protocols are
pluggable and that plugins can provide external function. But those are
not good reasons to put the whole Web services section under plugins.
Web services are not a plugin type and they always felt very
counter-intuitive there.
From the server administrtaion perspective, web services are similar to
scheduled tasks and therefore, they should live in the same section.
Using the external method for updating posts, the check for empty
subject/message content wasn't correct (disallowed the string '0'),
in addition to being impossible to set a posts message format
property to FORMAT_MOODLE (integer 0).
Modifies the 'string time to timestamp' behat trasformation to use
userdate() instead of date(). The userdate() method is generally used
throught Moodle for constructing formatted date strings and this change
will provide more consistency and prevent any potential behat failures.
Therefore, if the date format is defined in the given trasformation, it
has to be strftime compatible. Example:
'I should see "##tomorrow noon##%A, %d %B %Y, %I:%M %p##"'
Apart from adding new unit tests for covering new IMS OBv2.1 issuer,
some extra tests have been added to test existing issuers and confirm
they are still working as expected after the changes.
IMS OBv2.1 services have a registration endpoint to get client id
and secret.
This patch adds and implements the "register" method for the IMS
Open Badges Connect discovery system, to get the proper client id
and secret values.
In order to make easier to create and maintain new OAuth2 services,
a couple of classes have been added:
- discovery\* contains methods related to the discovery system. Until
now, only OpenID Connect was supported by Moodle so all the code was
centralised in api.php.
With this patch, as IMS OBv2.1 has a different discovery system, a new
abstract class (base_definition) has been added to be called and let
every discovery system (OpenID Connect, IMS Badge Connect...)
implement their own methods.
- service\* . Instead of keep adding methods to the api, the service
namespace has been created to store all the standard issuer services.
An interface (issuer_interface) has been created and all the services
should implement it.
This patch creates the "IMS OBv2.1" and "Custom" services and moves
the methods for "Google" service because it uses the OpenID connect
discovery system.
I've gone over a few of the mofified files (those
which were showing warnings and errors to CiBoT. Some of them
have been fixed completely, while others only have fixed
for the lines belonging to this issue (lib/tests/moodlelib_test.php)
for example.
This applies the "whitelist" => "include" changes to all the core
phpunit_coverage_info occurrences, so core won't emit any deprecation
warning (see previous commit).
At the same time, modified a bunch of comments in coverage files
to be more readable/understandable.
Following the changes in the schema, from "whitelists" to "includes",
we are deprecating these two properties from phpunit_coverage_info:
- whitelistfolders => includelistfolders
- whitelistfiles => includelistfiles
They will continue working over the deprecation period but the init/util
scripts will throw some warnings about them being deprecated for 3.11 and
the way to replace them.
Standard 2y deprecation applies with final removed to happen @ MDL-71067
Basically switch to 2-space indenting and small tweaks to comments,
so both the main phpunit.xml file and also the individual components
ones, all them based in the dist one will be consistent.
This performs all the changes needed in the util generator to
produce XML files compliant with the new schema (see previous
commit for description of changes).
- All the occurrences in code of filter => coverage.
- All the occurrences in code of whitelist => include.
- Apply all the changes to comply with the new schema.
- Remove processUncoveredFilesFromWhitelist attribute, useless (false
is its default value, and now have another name).
- Move from 4-spaces indented XML to 2-spaces indented.
- Small linefeed tweaks to generate better-looking XML.
PHPUnit 9 comes with various changes in the XML
configuration file, namely:
- the old "filter" section has been renamed to "coverage".
- "whitelist" has been renamed to "include"
- "exclude" is not a child of "include" anymore, but of "coverage".
- "include" cannot have configuration attributes anymore, only
"coverage" can"
Visually it means that the old section (invented example):
```
<filter>
<whitelist attributes_may_go_here="xx">
<directory suffix=".php">classes</directory>
<directory suffix=".php">externallib.php</directory>
<exclude>
<directory suffix="_test.php">.</directory>
</exclude>
</whitelist>
</filter>
```
Now looks like:
```
<coverage attributes_may_go_here="xx">
<include>
<directory suffix=".php">classes</directory>
<directory suffix=".php">externallib.php</directory>
</include>
<exclude>
<directory suffix="_test.php">.</directory>
</exclude>
</filter>
```
So, switching to the new xsd so we can validate the remaining changes.
The current ->setMethods() has been silently (won't emit any
warning) in PHPUnit 9. And will stop working (current plans)
in PHPUnit 10.
Basically the now deprecated method has been split into:
- onlyMethods(): To point to existing methods in the mocked artifact.
- addMethods(): To point to non existing (yet) methods in the mocked
artifact.
In practice that means that all our current setMethods() calls can be
converted to onlyMethods() (existing) and done. The addMethods() is
mostly useful on development phases, not final testing.
Finally note that <null> isn't accepted anymore as parameter to
double all the methods. Instead empty array [] must be used.
Link: https://github.com/sebastianbergmann/phpunit/issues/3770
The methods assertContains() and assertNotContains() now perform
strict (type and value) comparison, pretty much like assertSame()
does.
A couple of new assertContainsEquals() and assertNotContainsEquals()
methods have been created to provide old (non-strict) behavior, pretty
much like assertEquals() do.
Apart from replacing the calls needing a relaxed comparison to those
new methods, there are also a couple of alternative, about how to
fix this, depending of every case:
- If the test is making any array_values() conversion, then it's better
to remove that conversion and use assertArrayHasKey(), that is not
strict.
- Sometimes if may be also possible to, simply, cast the expectation
to the exact type coming in the array. I've not applied this technique
to any of the cases in core.
Link: https://github.com/sebastianbergmann/phpunit/issues/3426
Mocke at() matcher is being deprecated with phpunit9 and
will be removed with phpunit10.
Source: https://github.com/sebastianbergmann/phpunit/issues/4297
Luckily we are using those deprecated matchers only in completionlib
tests, so there aren't many cases to modify. Now, we are using
supported matchers (once, exactly, never...) and the tests have
been reorganised to better represent the expected behavior (how
many times stuff is called, with which params and return values).
Custom autoloaders are deprecated with PHPUnit 9 and will be removed
with PHPUnit 10.
Since PHPUnit 8.5 custom autoloaders don't do much because that
version removed the ability to launch unit tests by class name
and that's exactly the reason we had a custom autoloader (to map
class names to files within our tests). See MDL-67673 about
when direct use of classes was deprecated (8.5), now removed (9.5).
So, as far as it's unused, removing it now, test still can be
selectively using any of:
- a relative path to file (although there are some restrictions comming
with PHPUnit 9, see https://github.com/sebastianbergmann/phpunit/issues/4105
- using --filter, to point to any classname[::method]
- using --testsuite to run a complete suite
- using --config to point to custom components.
Also, commented out the lib/ajax/tests directory because it doesn't
exist / is empty and PHPUnit 9 emits error when a configured test
directory does not exist. See
https://github.com/sebastianbergmann/phpunit/issues/4493.
Alternative was to completely remove the configuration line, but
decided to keep it around in case some day we want to add some
test there.
Used by our custom assertTag() and assertNotTag() assertions, that some
day we should deprecate... the loading of XML content for further
processing has been moved to new classes within the PHPUnit utils. We
are just following the move here.
With stricter typed param checks in php73, the 7th param of the
getMockForAbstractClass() must be array, so previous code defaulting
to null now throws a TypeError.
In PHP 9.1, the use of expectException(PHPUnit\Framework\Error\*) has
been deprecated, that is, when a Notice/Warning/Error/Deprecated
problem is reported. Instead, these new assertions must be used:
- expectDeprecation() for E_DEPRECATED and E_USER_DEPRECATED
- expectNotice() for E_NOTICE, E_USER_NOTICE, and E_STRICT
- expectWarning() for E_WARNING and E_USER_WARNING
- expectError() for everything else
More info:
https://github.com/sebastianbergmann/phpunit/blob/9.0.0/ChangeLog-9.0.mdhttps://github.com/sebastianbergmann/phpunit/issues/3775
Regexp to find all them:
ag 'expectException.*(Notice|Warning|Error|Deprecated)
In PHPUnit 9.1, the following regexp-related assertions
have been deprecated and there are new alternatives for
all them:
- assertRegExp() -> assertMatchesRegularExpression()
- assertNotRegExp() -> assertDoesNotMatchRegularExpression()
This is about to, simply, move all cases to the new alternatives.
Source: https://github.com/sebastianbergmann/phpunit/blob/9.1.0/ChangeLog-9.1.md
Regexp to find all them:
ag 'assertRegExp|assertNotRegExp' -li
In PHPUnit 9.1, the following file-related assertions
have been deprecated and there are new alternatives for
all them:
- assertNotIsReadable() -> assertIsNotReadable()
- assertNotIsWritable() -> assertIsNotWritable()
- assertDirectoryNotExists() -> assertDirectoryDoesNotExist()
- assertDirectoryNotIsReadable()-> assertDirectoryIsNotReadable()
- assertDirectoryNotIsWritable()-> assertDirectoryIsNotWritable()
- assertFileNotExists() -> assertFileDoesNotExist()
- assertFileNotIsReadable() -> assertFileIsNotReadable()
- assertFileNotIsWritable() -> assertFileIsNotWritable()
This is about to, simply, move all cases to the new alternatives.
Source: https://github.com/sebastianbergmann/phpunit/blob/9.1.0/ChangeLog-9.1.md
Regexp to find all them:
ag 'assertNotIsReadable|assertNotIsWritable|assertDirectoryNotExists|\
assertDirectoryNotIsReadable|assertDirectoryNotIsWritable|\
assertFileNotExists|assertFileNotIsReadable|assertFileNotIsWritable'
Manage backpacks page wasn't displaying the navigation and administration
menus when using classic theme (it's unrelated to changes done here).
This patch fixes this unexpected behaviour (raised thanks to behat tests).
Since cm_info::customdata can be of any type, we need to cast it to an
array first before checking for custom completion rules. Otherwise,
an exception can be thrown (e.g. customdata has been set as an stdClass)
In all cases changes have been kept to a minimum while not making
the code completely horrible. For example, there are many instances
where it would probably be better to rewrite a query entirely, but
I have not done that (in order to reduce the risk of changes).
The $CFG->badges_site_backpack setting has been completely removed
because it's not required anymore. From now, the primary site
backpack will be the first one in the "Manage backpacks" list (so,
the one with lower sortorder value).
Before removing $CFG->badges_site_backpack setting, admins should be
able to re-order the existing site-backpacks (because then, the first
one will be treated as the default one).
This patch adds the sort order feature to the backpack list.
Setting "External backpack connection" has been moved from "Backpack
settings" to "Badges settings" in order to make easier to find and
configure it.
Some extra tests have been added to cover existing behaviour and
confirm is still working as expected.
When hovering over table rows with the `dimmed_text` class we
should consistently set the text/link colour.
Co-Authored-By: Mikel Martín <mikel@moodle.com>
Use the custom completion implementation for mod_choice to test
completion_info::get_state() to cover the case where the completion
state is being determined from the custom completion condition.
* Update completion_info::get_data() to add other completion
information from a new method called get_other_cm_completion_data().
This allows the storage of the completion statuses of the following
completion rules to completion_info objects:
- 'Students must receive a grade' completion rule.
- Any custom completion rule defined by an activity.
This allows detailed completion information to be fetched for course
modules.
It also allows custom completion statuses to be cached which will help
reduce DB queries when fetching completion statuses.
* Update update_state() to fetch overall completion state from the
module's activity_custom_completion implementation. Falls back to the
*_get_completion_state() callback function.
* Update internal_set_data() to include the other cm completion data
in the updated cache data for the module instance.
Move the current logic for determining the completion status for the
"Student must receive grade" completion rule to a function so it cann
be reused.
Unit test included.
* Unit tests for completion_info::get_data() and
completion_info::internal_get_state are mocked which causes failures
with the new implementation. It's more straightforward and realistic
to generate real course and modules to test these methods.
Tests cover
- get_overall_completion_state()
- is_available()
- validate_rule()
Tests don't cover
- methods that rely on static methods such as:
- is_defined()
- static methods in the class because they can't be mocked
- abstract methods that can be tested better by the plugins
that extend activity_custom_completion such as:
- get_state()
- get_defined_custom_rules()
- get_custom_rule_descriptions()
A number of behat fails have been creeping in with the customfield code
as behat is moving onto the next step before the current step finishes.
This change introduces some additional pending JS tracking to prevent
this.
Currently admin_setting_configselect has lazy-loading support via a
callback function (so you don't have to make pointless single-use
classes for each unusual setting), but this is not present in other
similar types.
This commit adds identical support to
admin_setting_configmulticheckbox.
Some of the current behat tests were wrong but not failing because of the bug in 'should not exist' step.
Once the bug is fixed, we must fix also wrong behats.
This text is not directly output to final users, it is used just in the textarea in both the site and the app.
Applying format, case issues like adding glossary links etc...
After uninstalling a plugin, the translated strings remain in
tool_customlang table, throwing an exception when trying to localise
any strings. Currently there is no mechanism to
clean up customlang tables and files during the uninstall process,
so with this patch the invalid components will be ignored.
Sortorder field in badge_external_backpack was updated with a wrong
value. This patch fixes this behaviour and adds sortorder as
sorting criteria for displaying the list of site backpacks.
In the phpunit method reviewed for covering this, a couple of
assertEquals() calls have been changed too to put the expected value
first.
It was deprecated in php72 and now it's gone.
Have used this regexp to find all the uses in core:
ag set_error_handler | uniq
And then checked all them manually, that parameter was not
being used in the 3 methods where we are removing it.
Implementing these functions as final does not prevent the object to be
cloned or serialized, so the comment to them was wrong.
However having these functions not public breaks in PHP 8.0.
It still does not make sense to clone or serialise file_system or external_settings classes
but there are a lot of classes in moodle that should not be cloned, no reason to have
a specific solution here.
PHP8.0 shows error that filter function accepts value by reference
The nested filtering was never working but also this function was never called on a tree with sub-levels,
so this logic is now removed
Set max-height on each container section rather than the container
itself, so that each can respond to the scrollBottom event within
the lazy loading module (which handles the overflow itself).
If a developer has debugging on then they will recieve all debugging
messages for locking whether wanted or not. We already have a setting
to display these messages.
pdftoppm (from poppler-utils package) is several orders of
magnitude quicker than ghostscript extracting PNG images
from PDF documents.
We add support for this tool and use it whenever set up,
or using gs as before if missing.
Revisited code with peer review comments. In particular:
1. Make default path for pdftoppm empty string.
2. Fix a typo on the method description for get_gs_command_for_image.
3. Added more information why is useful pdftoppm into string pathtopdftoppm_help.
4. Check that path for pdftoppm is executable to prevent errors. Otherwise, use gs.
High-resolution image for a badge (f3) should be sent to external
backpacks through JSON and WS methods.
Besides, when a badge is cloned, the high-resolution image should be
used to avoid the copied looks blurry.
The "Private files" link does not exist in the site navigation when
using the classic theme, so it is not possible to test the dedicated
page via Behat at this time.
As discussed in MDLSITE-6397, all Moodle code files should end with
single Unix-style end of line character. Files created and edited
through the XMLDB editor should follow this rule, too.
The filepicker wasn't displayed fot the H5P Atto button when a text
area database field was created because some options were missing
in the textarea field class.
* Old backups might have hidepicture defined and set to 1 in which case
we must make sure to exclude the group picture for these groups from
being restored.
Plus additional amendments to Fred's original commit:
1. Updating the version numbers
2. Merging the original two-step upgrade below into one for simplicity:
1. Deleting the pictures for groups with hidepicture set to 1; and
2. Dropping the hidepicture field itself.
3. Converted array() usages to the short syntax [].
If the drawer toggle button was clicked in quick succession to close
and then re-open, then the delayed Aria.hide method would be called
which hid the drawer content when it finished opening.
File badges/oauth2callback.php has been removed and now badges
API will use admin/oauth2callback.php because it makes no sense having
it twice.
Credits go to Andrew Nicols for raising it!
Adds behat support for selecting date and time from a datetime selector
element. The passed values should represent a textual date and time
description wrapped in '##' (e.g. '##first day of January 2020 08:00##',
'##1 Jan 2020 10:30##'). Also, the value 'disabled' is valid and can be
used to disable the datetime selector element.
Adds behat support for selecting a date from the date selector element.
The passed values should represent a textual date description wrapped
in '##' (e.g. '##first day of January 2020##', '##1 Jan 2020##'). Also,
the value 'disabled' is valid and can be used to disable the date
selector element.
Moves the logic from guess_type() to a separate protected method
get_field_instance_for_element(). This would be quite useful for
form field classes as they can now use this method when there is
a need to determine the type of a given node element.
Includes the data-fieldtype attribute to the checkbox inline form
element. This element does not extend the parent inline element
template due to its specific nature and therefore this attribute was
missing.
On behat, we have to use a more specific test step if we are looking
for a "Go" button on a modal. The reason is that the html of the modal
content is put after the html of the "Go to top" button, so behat
may find the "Go to top" button if we just look for "Go".
WAVE considers any text with the size of 10px or below really small.
Increasing the font size of the navbar counter by 1px should not make a
big negative impact on the design, but improves accessibility.
In MDL-63040, it was decided that the page header for the dashboad page
should be removed.
However, Based on Success Criteria 1.3.1 and 2.4.6 from the WebAIM's
WCAG 2 Checklist, all pages should have an <h1> element.
So I put an sr-only heading on the dashboard and all other pages, where
no page header is shown.
Changed the element selector to get all array elements,
whether named or unnamed. The previus selector was only
working for elements that thir names were like name="something[]".
It was not working for name="something[a]" elements.
This is not related to the issue but I noticed it while trying to debug
it.
Back when this line was introduced in 9c140a681e, the ZipArchive had
not yet exposed this flag as a constant. It was added later with PHP
7.0.8 and we can switch to using it now.
The original implementation was based on ZipArchive::getStream() which
turns out to be very slow and if the archive contains many files, the
unzipping performance is very slow.
The patch changes the implementation to use ZipArchive::extractTo()
unless the extracted entry path contains a folder name ending with dot
(such as some/path./to/file.txt). There is a known upstream bug in the
PHP ZIP extension #77214 (also #74619 and #69477) so that we fall back
to keep using the stream in those cases.
* When a label is passed to an (advanced) checkbox and no text
was provided, the label will be displayed on the right of the the
checkbox element. So there's no point in passing an empty string for
the label just to render the text to the right of the checkbox.
* When a label is passed to an (advanced) checkbox and no text
was provided, the label will be displayed on the right of the the
checkbox element. So there's no point in passing an empty string for
the label just to render the text to the right of the checkbox.
* The aria-describedby attribute can be a list of element IDs that
describe the element. On form validation, the ID of the error message
container is added to this attribute which may already be containing
another ID. So we need to properly add/remove the error message ID
so that we don't delete any existing ID(s) in the aria-describedby
attribute during form validation.
Save comment in the assignment when the Save changes
or Save and show next button are clicked at the bottom
of the page.
Signed-off-by: Sujith Haridasan <sujith@moodle.com>
Uses of the httpsreplace and customlang tools are known to be extremely
slow.
In our CI infrastructure on slwoer DBs, the customlang tool can take
over 90 seconds to load the page, whilst the httpsreplace tool can take
up to about 60 seconds.
These changes set appropriate timeout factors to increase the timeout
accordingly.
Also, change:
get_real_timeout(30000)
to:
get_real_timeout(30) * 1000
because the original functions are defined to use seconds, and
having the 1000 around will help us detecting cases in the case
of a hypothetical review of uses.
Displaying a default group picture in forum posts leads to confusion,
so better to not display a default group picture when the group
picture is not set for the group.
Also, change:
get_real_timeout(30000)
to:
get_real_timeout(30) * 1000
because the original functions are defined to use seconds, and
having the 1000 around will help us detecting cases in the case
of a hypothetical review of uses.
Credits for Alexander Bias, for sharing the patch!
The content bank was designed for making easier to reuse content.
So "Link to the file" should be the default option when using the
contentbank repository, in order to let teachers modify their
content in the content bank and update automatically it everywhere
is used.
If the request to the OAuth 2 token endpoint fails show the response
body the endpoint returned with its HTTP status (when debug: DEVELOPER).
If no response is available show any error returned by Curl. Previously
none of this information was available making troubleshooting difficult.
If a token refresh fails in \core\oauth2\refresh_system_tokens_task an
exception is now thrown so that the result is shown as "Fail" on
admin/tasklogs.php?filter=core\oauth2\refresh_system_tokens_task
get_magic_quotes_gpc() always return false since PHP 5.3 so (1 == get_magic_quotes_gpc()) is never verified.
This allows me to safetly delete what I dropped out.
An case was found where the webdriver stop() call could fail in an
AfterScenario hook, leading to a complete rerun if no other errors were
experienced.
- Update fetch_dropbox_data to allow different result nodes and version
- Update search to the new URL with new params
- Update the get entities function to account for the change in structure.
The emoji categories were modified in the newer versions of the
emoji-data library. Therefore, the category list in $categorysortorder
needs to be updated with the correct category names in order to properly
sort the categories in the generated emoji data file.
The emoji data generator script is enabled to detect inconsistencies
between the emoji category names used in the emoji-data library and the
ones used in Moodle. These inconsistencies should be fixed, as soon as
they are detected upon library upgrade.
Switch the order of operations performed when collating list of paths
from which a user can include attachments.
First collect all normalised/absolute paths then filter empty entries,
which fixes an issue where $CFG->localrequestdir could be defined but
not exist. This would lead to an empty string being passed to strpos
which triggered a PHP warning.
Co-authored-by: Peter Burnett <peterburnett@catalyst-au.net>
This ensures that the page reloads if the student uses
the Back or Forwards buttons within an attempt. This
avoids questions being in a stale state, or
the timer showing the wrong time.
Thanks to Jake Dallimore and Russell Boyatt for suggestions
which lead to this fix.
This fixes an issue on those systems that don't fully support unicode
characters within zip structures, which subsequently led to broken
links when browsing downloaded course content.
I can't expand this one in a sane fashion but the older JS is doing some
weird and wonderful things with focus.
The only way I could get this to behave correctly was to move away from
the field, move back to it, and then move away again. Other combinations
failed in different browsers for different reasons.
Expand all fieldsets rather than individual fieldsets.
This can fail in some situations where the page moves as the first
fieldset opens and HTML transitions take effect. This causes subsequent
fieldsets to move after WebDriver has calculated the click target.
These steps were creating content manually in an unnecessary fashion. I
had to debug some issues with features in these areas so took the time
to update them to make use of data generators, which are significantly
faster.
Behat should not trigger Synthetic browser events. It is incorrect to do
so as the UI should be used to trigger events correctly.
The W3C WebDriver specification explicitly states when and where these
events will be triggered from (the browser) and therefore there is no
ambiguity and no need to synthetically trigger them from Behat.
The W3C WebDriver leads to faster operation of individual steps which
highlights where WebDriver is moving to a new step faster than the
browser can process it.
The solution here is to use the `execute` function to interact with the
browsers as this makes use of the various wait and check steps.
These steps were completely pointless and were failing with the W3C
Webdriver becuase focus was never on the Update Profile button in the
first place.
At a minimum we should remove the Focus change step as it not valid, but
the following step is also pointless.
On Firefox there can be some issues with ensuring that all nodes are
visible in order to click them.
This is likely an upstream geckodriver bug and not one that can be
easily addressed across all of Moodle.
In this instance the "Next" button is just outside the Visible Window,
but it is in an iFrame which is fully visible. In this situation
Firefox is unable to click the button but does not error.
The W3C Specification does not require that a NodeElement be in the
current viewport before triggering a mouseOver. As a result the
mouseOver will generate an Exception because the element is not in the
viewport, and the X/Y co-ordinates are invalid.
To handle this the node is scrolled into view.
Normalise switching of window contexts by setting the main window name
to a null.
Where the window has no name (null/empty), then the root context is
selected.
This change introduces a new function to execute Javascript directly on
a node.
This should not, ordinarily, be used directly by steps, but may be
required in other parts of the Behat interaction.
The PHP WebDriver Mink Driver does not necessarily require Selenium. It
supports Selenium 3 upwards, and can be used directly with ChromeDriver,
EdgeDriver, SafariDriver, GeckoDriver, and others.
It therefore uses a slightly different configuraitno path.
This commit updates core Behat features to make use of the
php-webdriver/webdriver library instead of the legacy Instaclick
library.
Most of these changes relate to use of features of WebDriver which we
are required to use directly rather than through the Mink Driver.
This commit updates the composer dependencies to make use of the updated
version of moodlehq/moodle-behat-extension which depends upon the
php-webdriver/webdriver and relevant Mink WebDriver.
eg in Apache you could capture 403 errors and funnel them to this
error handler script and also transform them to 404s:
ErrorDocument 403 /error/index.php?code=404
When current language doesn't have accompanying videojs language file
we should fallback to displaying the player in English, otherwise the
web service returns empty and a Javascript exception is triggered.
This just deletes all the upgrade steps previous to 3.6.0. Some
small adjustments, like tweaking globals can also be applied
when needed.
Also includes an upgrade step to prevent upgrading from any
version < 2018120300 (v3.6.0) as anti-cheating measure.
Note that in this case, there wasn't any case of upgradelib
functions being used, hence we haven't to deprecate/remove
anything in codebase. When there is such a need, that is done
in separate commits (one for each function) and documented here.
See MDL-65809 commits for an example removing/deprecating a
good number of functions.
Upgrade step that resets the 'Data modification API' (convert_data)
setting to its default value if this setting is currently configured
to use a file located within the $CFG->dataroot directory.
Prevents configuring the 'Data modification API' (convert_data) setting
to use files located within the $CFG->dataroot directory as it exposes
the site to security risks.
This patch adds a solution for Postgres that allows overriding IN
statement limit of 16bit (65535 values) by wrapping items in VALUES
list. This only takes place in number of items exceeds 65535, otherwise
parent class get_in_or_equal() function is used.
If the activity uses a referenced H5P file from the content bank,
a link for displaying this specific content will be shown into the H5P
activity form instead of the generic one to the main content bank
page.
* The fact that non-existing strings are returned in certain format
starting with the bracket, is kind of side debugging effect. The value
is really undefined and should not be relied upon.
* The string could actually exist and be worded so that its first
character is also the square bracket.
The form_autocomplete is essentially a custom element. Unfortunately the
`setValue()` function in Mink has undesired actions so it is necessary
to write our own handling for it.
The standard Mink `setValue()` function focuses the element, sets a
value, and then blurs the element. In the case of the autocomplete this
can cause the autocomplete suggestions list to be closed in some
situations. Instead of using the setValue we click, and type the value,
but do not immediately blur.
Fixed a regression caused by MDL-64194 resulting in Dashboard - Course
overview pagination not working from second page of courses onwards.
Course overview block should now do the following:
- Display no pagination controls when user has no courses
- Display no pagination controls when less than 12 courses to display
- Only display pagination controls up to the number of courses user is
enrolled in
- Work correctly when on a page greater than the second page.
There was a typo - missing underscore. As a result, the new refresh
token was never updated in the database. Depending on the issuer and
whether or when theyu invalidate issued refresh tokens, this might or
might not make the token refresh stop working.
Creates a webservice that returns the addable blocks in a given page.
The webservice expects the page context id, page type and page layout as
parameters and returns the block name and block title for each available
block.
Starting from PHP 7, most language errors throw Error exceptions. We
want to deal with those equally as with any other exception so we need
to catch all Throwables here.
There are serveral ways a quiz attempt can be submitted:
1. The student click the Submit and finish button. In this case,
no problem. We record the current time as the finish time
for the quiz attempt.
2. The student is activly working away at the quiz, and the
count-down timer reachers zero. In this case, we also record
the current time. Note that, if the server is under high load,
then this could well end up being a few seconds after the
theoretical end time, so you could have a quiz with a 30 minute
time limit, with an attempt that lasted 30:07. However, this
is just an accurate reflection of what happened, so should
be recorded like this.
3. If the student is offline when the time expires, then
(depending on the quiz settings) the attempt may be
automatically submitted by cron, but this will happen with
at least some delay (to prevent race conditions between cron
and a student working online) and if cron is running slow
on the server, it could be a lot later. Previously, this led
to, say, a 30 minute quiz where an attempt seemed to have
lasted 67 minutes, which confused people.
Now, in this situation, the finsh time for the quiz attempt is
recorded as the time when the time limit ran out. This is not
just less confusing for teachers looking at the quiz report,
it is also more accurate. That is the latest time at which
students could have made any changes to their responses.
* Set appropriate heading element IDs
* Replace usages of deprecated HTML tag acronym with abbr instead.
* Add a label for the export fields fieldset.
* Move the CSV delimiter list outside the radio group.
* Fix div under a label element and duplicate labels
for the checkboxes of the fields to be exported by showing the field
name and type together.
* Group the fields to be exported as a fieldset.
* Move unsupported fields into a separate list.
* Move export options into its own section and group the options as a
fieldset.
First working version, supports phpunit (using build matrix):
- php72 (lowest), running mysql.
- php74 (highest), running postgres.
Also verifies that the branch has been "gruntified" and there
isn't any missing change (build js/css files).
TODO: Verify the remaining checks currently in .travis.yml, namely:
- CITEST
- Add caching
- Better health-check for DB images.
- Support from the tracker (satus badges and enable check).
- Support from CiBoT (status and enable check).
- Consider moving both the common setup (git, composer...)
and the database (mysql, postgres) to own actions for
easier tweaking.
Behat does not play well with confirmation dialogues and in some
situations it does not wait for the page to reload before starting the
next step.
This change adds a pending_js call which is only resolved if the cancel
button was pressed, otherwise it remains in-progress until the page
reloads resetting the entire state of it.
This commit updates the following steps to use the
`the_attribute_of_should_be_set` step under the hood:
- the [element] [type] should be disabled
- the [element] [type] should be enabled
- the [element] [type] should be readonly
- the [element] [type] should not be readonly
This reduces unnecssary code duplication.
This commit promotes the Inplace Editable field to a first-class form
element by introducing a new partial selector for inplace editable
fields, and teaching the field manager how to recognise these, then
introducing a new field type which can handle setting values for this
field.
Behat form fields are implemented in a way completely isolated from the
rest of the Behat Context system. Whereas regular step definitions have
access to execute steps, to call `find`, check for JS running, and other
related functionality, the Moodle implementation of a field type does
not have any access to this.
By moving the core functionality of behat_base to a new trait, and the
constants to a new interface, the functionality can also be used in
behat form fields in the same way as elsewhere.
This is similar to change made in MDL-69136 to allow an already-fetched
NodeElement to be provided to the get_node_in_container() function and
makes it easier to be deterministic when writing steps.
By default only lowest php version will be executed, and only pgsql.
This default behavior can be changed with a new variable:
MOODLE_PHP = [all]
MOODLE_DATABASE = [pgsql | mysqli | all]
The 'edit_instance_validation()' method checks for existing meta enrolment instances.
The fix replaces DB queries in a loop for each course with a single query for all courses.
Also, a new testing method 'test_edit_instance_validation_with_existing_courses()'
was added to /enrol/meta/tests/plugin_test.php to test if the new implementation
returns an error in case of trying to save the already linked courses in the 'customint1' field.
In some browsers the ajax grade select does not properly update the
grade after setting values. The previous solution was to press the
[enter] key, but doing this with the new key type step opens the select
box again. This is what happens when a real user presses enter on the
select.
This is the last possible field in the report, so pressing the tab key
to move to the next gradable element does not work.
The solution uses a shift-tab to move the focus away to the previous
gradale element. In this case it must also be moved to an earlier step
because the previously selected value must be checked in the Then
section of the test and if it is selected then its value cannot be
checked.
Note: The ddmarker question type was previously getting the number of
keypresses wrong. This was because it was using both keyDown/keyUp, and
also keyPress. As a result each keypress was essentially happening two
times.
A previous change as part of this commitset was intended to move the
modal to the document.body, but where the modal had been destroyed it
was instead added back to the body.
Before this change when a modal was shown and an element on the page
was fullscreen the modal would be created behind it.
This change ensures that the modal will be inside an element that is
fullscreen so that it will be correctly displayed.
- Main version (version, release, branch and maturity), guaranteeing
we are diverging from previous stable branch and won't overlap any more.
- Backup release, to match new branch.
- CI configuration files (.travis.yml in this case), to match new branch.
For changes needed to START working with parallel branches, see MDL-69475
In YouTube Data API v3, the API call to fetch the videos related to a
certain category expects the category ID to be passed to the
videoCategoryId parameter, instead of the category name. Therefore, the
category config should store the category ID.
When we read the MUC configuration, a file which exists but is empty
will not error, but will cause all configuration to be empty.
We already perform an {{is_array()}} check on the {{$configuration}}
variable, but the default value for {{$configuration}} is an empty
array. In the case where the file exists, but is empty, no errors occur
when the file is loaded, and the initial {{$configuration}} value is
not overwritten, leading to the file being replaced with an empty copy.
Having mixed $data in badge exporters is causing some issues.
As all these exporters are using $data as an object, $data can be
converted to object in the constructor, to avoid errors and get
the expected behaviour always.
Ensure that there is always one active element in the list of selected
autocomplete elements.
Without this we have issues beacuse clicking on the link makes the first
one active if one is not already active, and this turns a click event
into a drag event, which means that it is not deleted.
Returning a value of `false` in an event handler has the effect of
calling event.preventDefault() and event.stopPropagation().
This is neither obvious, nor desirable in this situation.
The documented values that jQuery.attr() accepts are String, Number, or
null. For some reason, when we pass a Boolean value, the subsequent
click handler does not work in some situations.
Changing this to take a Number, and unsetting it when empty, resolves
this issue.
The accessibleChange custom interaction event was only listening for
blur and focus, however some OS/browser combinations do not emit these
events until the element is explicitly blurred. This is notably
different on Firefox on some Operating Systems.
Recent changes in MDL-68167 explicitly moved the user participants page
filter module to use the accessibleChange event, which means that the
selections are now only triggered on an explicit blur when using
Firefox. This highlight a bug whereby, when the mouse is used to make a
selection, the event is not triggered until the element is blurred.
This change modifies the accessibleChange event to ignore the `change`
event where it was triggered by the keyboard and where that keybaord
event was not a [return] or [escape] keypress, but to otherwise respect
the native change event.
The `blur` event does not bubble, but the `focusout` events are not
available in all supported versions of Firefox.
Rather than switching event, this patch using event capture to
effectively achieve the same result and bubble the event up through the
DOM to the delegated listener.
There should be no functional change with this patch, except to support
Firefox fully.
version = 2020110900 release version
requires= 2020110300 current beta+ (week6roll1) version
Note that, because we are under parallel development period,
this is being done in the branch that is going to be released
(MOODLE_310_STABLE already existing) for Moodle 3.10.0 and
not in master, that is the one getting the bump under normal
(non-parallel) periods.
When 2 or more backpack were created without credentials,
a "Duplicate key value violates unique constraint" error
was raised because externalbackpackid was not taking the
correct value.
Other improvements have been done to the code too in order
to make it more readable.
Also, ensure that remove_dir() only processes directories,
because sometimes it was being called by shutdown managers
with files, leading to PHP warnings.
Co-authored-by: Eloy Lafuente (stronk7) <stronk7@moodle.org>
Co-authored-by: Jun Pataleta <jun@moodle.com>
With MariaDB 10.2.7 the driver was incorrectly returning '' (empty
string) as default for all the columns having null (meaning, no
default).
Also, cover the case in unit tests, it seems that we were already
testing other defaults but not the null case.
The content of the autocomplete selection div is purged by javascript
so we have to move its label outside of it. aria-labeledby is the way
to go.
Each time the suggestion box is expanded, autocomplete js renders the
autocomplete suggestions template. So the label cannot be outside the
suggestions ul.
* With the removal of the label element, Behat now has to click on the
answer text. In order to achieve this, the custom partial named selector
"qtype_multichoice > Answer" was now added and have replaced the
instances where the label of the answer's checkbox/radio button is
being clicked.
* Discard the use of the label element in order to be able to render
multiple choice answers as they are and have these act as the radio
button/checkbox' label through the aria-labelledby attribute.
* New JS module qtype_multichoice/answers that listens for click events
on the answer text container and selects the appropriate answer radio
button/checkbox.
- Add help for 'payment account' field in the enrol instance form
- Remove MOODLE_INTERNALs when not necessary
- Add $userid to deliver_order
- Check if provider classes implement the provider interface
- Rename get_cost to get_payable
- get_payable returns payable object
- Improve registerEventListeners and added init
- Rename payment\provider to payment\service_provider
Also create the infrastructure to let components know when they have to
deliver what they sold. We are going to use namespace functions instead
of traditional callbacks.
It was preventing JS click event handlers being reached on some
elements, such as course admin items in Classic
(eg download course content for teachers).
There was an error when creating the unique index. Apart from that,
the code has been reorganised to copy data in the new structure only
if password field hasn't been removed.
There are some unexpected differences between 310 and master,
better reconcile them now. Note this is not 100% critical but,
as far as nothing exclusively for 4.0 has landed to master yet
only branch/version differences should exist.
This includes:
- Breaking changes (return void + PHP 7.1 min requirement).
- Deprecations and removals.
- Specifically DbUnit and its replacement.
- Printer showing how to rerun tests removed.
That custom printer was using some hacks to be able to intercept
configuration switches, reflection and other tricks to be able to
print the:
To re-run:
vendor/bin/phpunit --verbose "core_setuplib_testcase" lib/tests/setuplib_test.php
line on every failed/exceptional/skipped test. After some internal
discussion it was agreed that the normal phpunit output:
1) core_setuplib_testcase::test_localcachedir
Time is lower that allowed start value
Failed asserting that 1601976686 is equal to 1601976687 or is greater than 1601976687.
/var/www/html/lib/phpunit/classes/advanced_testcase.php:446
/var/www/html/lib/tests/setuplib_test.php:170
/var/www/html/lib/phpunit/classes/advanced_testcase.php:80
has already all the information at hand about how to rerun a test:
- vendor/bin/phpunit lib/tests/setuplib_test.php
- vendor/bin/phpunit --filter core_setuplib_testcase::test_localcachedir
- vendor/bin/phpunit --filter ::test_localcachedir
- vendor/bin/phpunit --testsuite componentname_testsuite
- vendor/bin/phpunit --config <<compoenent directory>>
- use --cache-result to get failed tests rerun with ease.
- ...
So better, let's use standard phpunit output and done. Also, note that,
with the upgrade to phpunit 8.5, the printer was not working correctly
any more, causing some switches, like --verbose ... to be ignored. Sure
it could have been fixed but, as commented above, no real need for all
that "parapheranlia" to print the rerun information.
Apply the standard deprecation procedure to the old APIs, to
be removed in Moodle 4.2 (MDL-69882). Existing unit tests
cover the deprecation debugging.
Originally MDL-64600
In charge of taking over the, now removed/archived, phpunit/dbunit
package. It supports loading of CSV/XML files and strings and
PHP arrays, allowing to send the loaded information to database.
Perform some basic controls about the consistency of information,
surely not super-exhaustive but fair enough.
100% covered with unit test.
Planned like an "interim" replacement for phpunit/dbunit uses in
core that, ideally, should be moved to generators stuff at some
point.
Note, expect a few tests in core to fail with this commit, I've
changed some fixtures around. Next commit will fix existing uses.
Originally MDL-64600
It can be enabled if desired from CLI using the --cache-result
option (or modifying the phpunit.xml cacheResult to true).
That will create the .phpunit.result.cache files that,
later, can be used to easily repeat failed (defects) tests or
run the slow ones first and more. See the --order-by option
to know more about all the available criteria. For example:
./vendor/bin/phpunit --order-by=defects --stop-on-defect
(will run all the failed ones first, stopping on problem, useful
for TDD and other scenarios, not so much for complete runs or CI).
Also, added to .gitignore so they won't be committed ever.
The optional parameters of assertEquals() and assertNotEquals()
are deprecated in PHPUnit 8 (to be removed in PHPUnit 9):
- delta => use assertEqualsWithDelta()
- canonicalize => use assertEqualsCanonicalizing()
- ignoreCase => use assertEqualsIgnoringCase
- maxDepth => removed without replacement.
More info @ https://github.com/sebastianbergmann/phpunit/issues/3341
Initial search done with:
ag 'assert(Not)?Equals\(.*,.*,' --php
Then, running tests and fixing remaining cases.
Both assertContains() and assertNotContains() are deprecated in PHPUnit 8
for operations on strings. Also the optional case parameter is. All uses
must be changed to one of:
- assertStringContainsString()
- assertStringContainsStringIgnoringCase()
- assertStringNotContainsString()
- assertStringNotContainsStringIgnoringCase()
More info: https://github.com/sebastianbergmann/phpunit/issues/3422
Regexp to find all uses:
ag 'assert(Not)?Contains\('
With PHPUnit 8 a good number of assertions, all them related with
operations on non-public attributes have been deprecated. And will
be removed with PHPUnit 9.
The main point is that unit tests shouldn't be testing non-public
APIs (good practice) and those assertions were an error originally.
See https://github.com/sebastianbergmann/phpunit/issues/3338 for
the complete list and other details.
When possible (the attributes being checked are public), the change
is simple, just switching to normal assertions.
When the attributes are not public we need to find a workaround
to be able to test the same using public APIs, or use Reflection,
or remove the tests.
For the records, this is the regexp used to find all the cases:
ag '>(assertAttribute|attribute\(|readAttributte|getStaticAttribute| \
getObjectAttribute)' -G "test.php"
While this is not strictly required, because removal will
happen in PHPUnit 9.0, we are already getting rid of all
uses in core.
From release notes:https://phpunit.de/announcements/phpunit-8.html
The annotations `@expectedException`, `@expectedExceptionCode`,
`@expectedExceptionMessage`, and `@expectedExceptionMessageRegExp`
are now deprecated.
Using these annotations will trigger a deprecation warning
in PHPUnit 8 and in PHPUnit 9 these annotations will be removed.
Also, all uses of expectExceptionMessageRegExp() has been moved
to expectExceptionMessageMatches(). See https://github.com/sebastianbergmann/phpunit/issues/3957
TODO: Various weirdness found while doing the changes with these tests:
- vendor/bin/phpunit lib/tests/exporter_test.php (created MDL-69700)
- vendor/bin/phpunit competency/tests/external_test.php (same issue than prev one)
- vendor/bin/phpunit question/engine/tests/questionengine_test.php (created MDL-69624)
- vendor/bin/phpunit lib/tests/event_test.php (created MDL-69688)
While this is not strictly required, because removal will
happen in PHPUnit 9.0, we are already getting rid of all
uses in core.
From release notes:https://phpunit.de/announcements/phpunit-8.html
assertArraySubset() is deprecated and will be removed in PHPUnit 9.
While this is not strictly required, because removal will
happen in PHPUnit 9.0, we are already getting rid of all
uses in core.
From release notes:https://phpunit.de/announcements/phpunit-8.html
assertInternalType() is deprecated and will be removed in
PHPUnit 9. Refactor your test to use assertIsArray(), assertIsBool(),
assertIsFloat(), assertIsInt(), assertIsNumeric(), assertIsObject(),
assertIsResource(), assertIsString(), assertIsScalar(),
assertIsCallable(), or assertIsIterable() instead.
Constraints are now declared final, so we cannot extend them anymore.
Hence, instead of extending PHPUnit\Framework\Constraint\IsEqual we
are just wrapping it into our constraint.
All the setup/teardown/pre/post/conditions template methods
now are required to return void. This was warned with phpunit 7
and now is enforced.
At the same time, fix a few wrong function names,
provider data and param types, return statements...
This makes everything to break, but it's what we'll be fixing
in the next commits, heh. :-)
Generated with php72 (min version supported), following the
instructions @ https://docs.moodle.org/dev/Composer
Before this patch strings that were declared over multiple lines in
a template would not be parsed correctly.
{{#str}}value,
mod_forum{{/str}}
would have been been parsed to get_string('value', 'core')
{{#str}}
value, mod_forum
{{/str}}
would have been been parsed to get_string('', 'core')
* Lighten info container.
* Fix colour contrast for the question container by
- Lightening the background a bit
- Darkening the text colour a bit
- Making sure the link colour has good colour contrast against the
question text.
* Fix colour contrast for the comments container by
- Lightening the text colour a bit
- Darkening the link colour a bit for better colour contrast against
the feedback text.
The current red, blue, and green colours achieve the minimum colour
contrast against a white background. However, when they're against
coloured background, text with these colours will easily fail the
4.5:1 contrast ratio requirement for Level AA.
The updated colours now have the following contrast ratio against a
white background:
* Blue - #0f6fc5 (5.12:1, from 4.58:1)
* Green - #357a32 (5.27:1, from 4.63:1)
* Red - #ca3120 (5.29:1, from 4.59:1)
The current red, blue, and green colours achieve the minimum colour
contrast against a white background. However, when they're against
coloured background, text with these colours will easily fail the
4.5:1 contrast ratio requirement for Level AA.
The updated colours now have the following contrast ratio against a
white background:
* Blue - #0f6fc5 (5.12:1, from 4.58:1)
* Green - #357a32 (5.27:1, from 4.63:1)
* Red - #ca3120 (5.29:1, from 4.59:1)
Makes sure the guest user context exists when fetching
the recommended modules. If the context does not exist
an exception is being thrown describing the problem.
Without this, there's a code path that under the right circumstances is hit and
results in a query with `FROM (mdl_event) ev` in it - which is not valid syntax
under postgres.
* Deprecated serialise_tool_proxy() as no longer used.
* Use tool_proxy_return_structure for each proxy.
* Added tests for getting all proxies and orphaned proxies.
The remove_course_contents function did not delete data from
course_module_completion table in most cases, resulting in large
quantities of orphaned data.
The new API works on normal databases (by deleting data based on the
subquery) and also on MySQL (by deleting the data using a weird join
on the subquery).
Before there is a conversation, messages between users are sent
using core_message_send_instant_messages instead of
core_message_send_messages_to_conversation.
That method had an undefined $errormessage variable, ultimately
causing the problem detected on message_post_message()/message_send()
to never been returned by the external.
This just adds the same error than already existing conversations do,
but returning it as warning, for BC, JS will show it.
In this commit, we also reorganised how the Mobile box is displayed in the user profile.
We are allowing now site admins to see the user last access time.
Pass current user object to post builder as argument, so that the permission to view timed post
will check with current user, who is viewing the posts instead of user who made that post.
For clients that wish to consume the original value of the custom
field (e.g. timestamps for date fields), rather than the formatted
version, add a new "valueraw" property to the returned data.
Request additional offline access, allowing the repository to sign in
automatically each time, including across different user sessions, until
either the refresh token is invalidated or the user logs out of the
repository.
Changes:
- The client can now be instantiated with the new mode enabled.
- A user refresh token will be stored if returned as part of the
authorization_code grant but only if the user is authenticated and only
if the mode is enabled.
- If the mode permits it, attempt to exchange a stored refresh token for
a new access token during login checks, allowing access to continue
across sessions without the need to re-grant consent every time.
- Purge all refresh tokens for a given user, issuer and scope on client
log out. This makes sure logout persists across logins.
When plugins are removed from disk, an exception is thrown when
calling component_callback_exists.
This exception should be catched here in order to load properly
the activity chooser (otherwise, the exception will be displayed
and no activity will appear).
The patch introduces a new admin_setting fiela type that can be used for
specifying comma separated list of countries. The field has inbuilt
validation so that only valid country codes can be inserted.
- The backup details page uses a table to show a sumary of the backup
content. Used role attribute to denote the tabular format of the
summary.
- The backup details page displays activity name next to each activity
icon. Therefore the icons are only decorative and do not need to have
any title or even alt text.
- Form labels should be associated with form controls. A div element is
not a form control.
- The from attribute of the form labels should be equal to the id
attribute of an element. Therefore, we first create a label and an
input elements and associate them to each other, and then pass them to
backup_detail_pair() when a label is needed.
Existing mod_hvp pluging has a renderer to let Moodle instances
alter styles, Javascript, semantics (fields in the editor) and
content: https://github.com/h5p/moodle-mod_hvp/blob/stable/renderer.php
The approach for core_h5p is exactly the same, to let people
to reuse their existing code.
Rename the string identifier to filetypesnotallowed, copy all the
existing translations in AMOS and deprecate the original string to be
eventually removed.
AMOS BEGIN
CPY [filetypesnotwhitelisted,core_form],[filetypesnotallowed,core_form]
AMOS END
The code has been using terms that were found exclusive and potentially
derogatory. Fixing the terms and making the variables and methods
consistent with the related admin settings.
The name of the placeholder has been found exclusive and potentially
derogatory. Also the name of variable that was used to populate the
placeholder value. That was not the intention.
If the allcountrycodes filter contains only invalid values, ignore the
whole filter setting and make get_list_of_countries() return the full
list of all known countries, rather than empty list.
The link class attribute is a leftover from 7258bf0b, which was mostly
removed in 3b2f691f. It no longer serves a purpose and interferes with
the element focus outline.
When editing a user and checking the 'Delete picture' checkbox to delete
the old picture and uploading a new one at a time, the new one didn't appear
in the user profile. This happened due to the 'if/else' statement
in the 'core_user::update_picture()' method which allowed either to delete
the old picture or to upload a new one.
The fix removes the 'else' part of the 'if/else' statement to force
checking for a new uploaded picture no matter if a user checked
the 'Delete picture' checkbox or not.
There were inconsistencies with the text shown on the submit button,
depending on the where the button was shown.
* Inline reply forum displayed general "Submit".
* Experimental nested discussion view reply form displayed "Post"
(post/core string) which acts as a noun in other contexts and cannot
be used as such in many non-English languages.
* Advanced reply form displayed "Post to forum".
The patch fixes this and starts to show "Post to forum" in all three
places consistently.
While working on this, two more issues were noticed and fixed:
* Inline reply textarea's title was hard-coded English "post" text.
Changed to "Message" to provide advisory information related to the
element.
* Buttons contained title attribute with the same value as the explicit
button text. This goes against accessibility guidelines.
Calls to core_get_course_contents can cause PHP warnings in the server
log if there are any URL instances in the course with sketchy URLs.
This is because url_export_contents incorrectly returns null instead
of an empty array if there is an error.
It was found out that in certain situations there may be some content
in to output handlers buffers. Let's allow that to happen given that:
* default handler's buffer is not sent as a part of the file and an
error is reported into the server log.
* other handlers' buffer is sent with the assumption that it is required
for the correct serving of the file.
The localisation must happen only after loading the list of licenses
from cache so that the name is displayed in the current user's preferred
language, not in the language of the user who initiated the cache
rebuild.
The autocomplete AJAX is identical to the version in tool_lp, but also
includes support for excluding suggestions (i.e. cohorts already
synced with the current learning plan).
If two different plugin types (e.g. theme, local) implement
override_webservice_execution callbacks, then all of them except
the 'last' plugin type do not work correctly.
New function get_total_content_size() is introduced in the stored_file
class. The puprose of this function is to calculate and return the
total size (in bytes) of the content of an archive file.
When a given user doesn't have the capability to "Include user data"
during course copying, freeze the form element rather than not adding
it at all.
This caused problems as the element was required before preceding with
the course copy.
Use $gray-600 for a darker text colour in the course select box.
This also makes the colour consistent with the view selector which
also uses $gray-600.
Increase the opacity for the sass variable $navbar-light-color a
little bit in order to achieve an acceptable colour contrast against
the white background.
The $component = 'mod_h5pactivity' will be passed always to the H5P
player because it's required for being able to call these activities
from outside using embed code. The xAPI implementation checks whether
the information should be tracked or not.
When the Moodle component is passed to the player, xAPI tracking
is enabled so, if the component implements required xAPI API methods
(for instance, mod_h5pactivity does), grades and responses are sent.
Adding this component to the embed code will help to add
the embed code from a mod_h5pactivity to any other place (such as label,
page or book), and being able to track the responses to the original
mod_h5pactivity.
- Screen-readers should not see 'clear my choice' when it is not visible
- 'clear my choice' option should only become visible when a choice is
selected
When the Moodle component is passed to the player, xAPI tracking
is enabled so, if the component implements required xAPI API methods
(for instance, mod_h5pactivity does), grades and responses are sent.
Adding this component to the embed code will help to add
the embed code from a mod_h5pactivity to any other place (such as label,
page or book), and being able to track the responses to the original
mod_h5pactivity.
As well as fixing the bug, I also rewrote the test to use
data providers, which should lead to more useful failure messages.
And, I moved the magic number we used as the float tolerence to
be a named constant.
* Introduce the "actionmenulinkclasses" block in the action_menu_link
template so other templates importing it can have the option to override
its classes.
* Have the action menu trigger template use the action menu link
template to reduce code duplication.
With the static function variable $instance, calls to
\action_menu_link::export_for_template() from its subclasses are stored
in different variables. This causes duplicate IDs when different
implementations of action menu links are rendered on the action menu
trigger/link template. (e.g. action_menu_link and
action_menu_link_secondary both rendered on the same page).
To make the incrementing uniform for the action_menu_link class and its
implementations, the $instance variable is moved out of the
export_to_template() method and is now made to a static class variable.
Links and normal text should have at least a colour contrast ratio
of 3:1. To achieve this, $body-color will be using the darker
shade of gray $gray-900 (#212529).
Links and normal text should have at least a colour contrast ratio
of 3:1. To achieve this, $body-color will be using the darker
shade of gray $gray-900 (#212529).
Darkened the fills of completion icons in order to meet the required
minimum contrast ratio of 3:1.
* Changed gray fills (#999) to #949494.
* Changed blue fills (#76A1F0) to #6393ee
* Changed green fills (#9C3) to #79a128
Thanks to Barbara Ramiro <barbara@moodle.com> for the help in
updating the SVG icons!
* Darkened the link and clickable icon colours inside coloured events.
* Added borders around event colour indicators inside the month-view of
the calendar for better contrast against the white background.
If a scheduled task which uses a 'R' field is picked, then the reset of
updated task times will not necessarily be correctly determined as the
randomisation is picked during reset.
This can lead to some random test failures.
Actively specifying a test which does not make use of the 'R' random
field time addresses this issue.
Before this change in a frozen context a user would be able to delete
their own comments.
After this change a user will only be able to delete their own comments
if they still have the capability to post new comments or the capability
to delete comments from any user.
Under parallel development, when something is worth being commented
in upgrade.txt notes, it will be applied to all the development branches
using the "earliest" one to be released. So, if something changes
in 3.10 (and also is applied to master), the upgrade.txt notes will
show 3.10 in both branches.
Of course, if something only goes to one dev branch (say master), then the
master version will be used in the notes (4.0 in this case).
This commit introduces the following new flags:
--no-composer-self-update Disable composer self-update of the
composer.phar utility
--no-composer-upgrade Disable upgrade of development
dependencies using Composer.
--disable-composer Disable both composer self-update and
composer upgrade
Both ldap or the DB can return information in a non-consistent
ordering leading to events to be generated in different order.
And current tests are, right now, assuming a given order.
Note this is a rare random, but it's happening, so better
fix it, see the issue for some more details.
So we just do the tests ordering immune, verifying that all the
expected events have been triggered and done. Irrespectively of their order.
Currently, PHP getimagesize method doesn't support SVG images.
As some features, such as badges, processs and optimise the images
before using them, a new filetype group has been created to exclude
SVG from there: optimised_image.
SVG can't be removed from web_image because then users won't be
able to add SVG images to their courses using labels, pages...
The external database authentication plugin (auth_db) can update the
external database if mapped fields are set to Update external: On
update. This change adds error handling in case this update fails.
Also removed the scope attribute from the mini-calendar table
because according to https://www.w3.org/TR/WCAG20-TECHS/H63.html
it was not needed:
Note 1: For simple tables that have the headers in the first row
or column then it is sufficient to simply use the TH elements without
scope.
1) Remove any floatval() casting. They are breaking / killing
.10 versions (converting them to .1). Since Moodle 2.0 all the
backup::RELEASE have been 100% numerical values.
2) Use version_compare() always to compare backup::RELEASE values.
They are always versions and the function is aware of versions
> .9, able to clean/ignore alpha chars... and everything else.
Note that I've also changed a couple of cases in formats (topics and
weeks) that were correct, but just added the same comment and used
the same version_compare() comparison parameters style, so all uses
in core are consistent (and safe to be copied out there).
On installation (or also phpunit/behat unit) some CFG variables
were being used (on setting validation) before being set.
So this commit just verifies they are set before using them. Note that,
strictly speaking, only one of them ($CFG->searchenginequeryonly)
required the extra check, but I think it's better to apply it to all
them, as a reference and in case validations are run in any other order.
- Also bump travis to get builds against 310_STABLE.
- And backup release to match 3.10
Note this doesn't aim to be a complete change-set, but just the
minimum to switch to 3-digit $branches and keep installation, checks
and tests running and passing.
2020-08-17 00:12:30 +02:00
9081 changed files with 320070 additions and 152223 deletions
//$temp->add(new admin_setting_configcheckbox('backup/backup_auto_messages', new lang_string('messages', 'message'), new lang_string('backupmessageshelp','message'), 0));
//$temp->add(new admin_setting_configcheckbox('backup/backup_auto_blogs', new lang_string('blogs', 'blog'), new lang_string('backupblogshelp','blog'), 0));
@@ -437,6 +522,9 @@ if ($hassiteconfig or has_any_capability($capabilities, $systemcontext)) {
$temp->add(newadmin_setting_configselect('lang',newlang_string('lang','admin'),newlang_string('configlang','admin'),current_language(),get_string_manager()->get_list_of_translations()));// $CFG->lang might be set in installer already, default en is in setup.php
Given Inavigateto"Plugins>Webservices>Managetokens"insiteadministration
And Ifollow"Add"
And Isetthefield"User"to"JoeBloggs"
Scenario: Add a token to user identified by name and then delete that token
Given Iloginas"admin"
And Iamonsitehomepage
And Inavigateto"Server>Webservices>Managetokens"insiteadministration
And Ipress"Createtoken"
And Isetthefield"User"to"Firstname1Lastname1"
And Isetthefield"Service"to"Moodlemobilewebservice"
And Isetthefield"IPrestriction"to"127.0.0.1"
When Ipress"Savechanges"
Then Ishouldsee"JoeBloggs"
And Ishouldsee"127.0.0.1"
And Ifollow"Delete"
Then Ishouldsee"Moodlemobilewebservice"inthe"Firstname1Lastname1""table_row"
And Ishouldsee"127.0.0.1"inthe"Firstname1Lastname1""table_row"
And Iclickon"Delete""link"inthe"Firstname1Lastname1""table_row"
And Ishouldsee"DoyoureallywanttodeletethiswebservicetokenforFirstname1Lastname1ontheserviceMoodlemobilewebservice?"
And Ipress"Delete"
And Ishouldnotsee"JoeBloggs"
And "Firstname1Lastname1""table_row"shouldnotexist
@javascript@skip_chrome_zerosize
Scenario: Tokens can be filtered by user and by service
Given thefollowing"core_webservice>Service"exists:
|name |Siteinformation |
|shortname |siteinfo |
|enabled |1 |
And thefollowing"core_webservice>Servicefunction"exists:
|service |siteinfo |
|functions |core_webservice_get_site_info |
And thefollowing"core_webservice>Tokens"exist:
|user |service |
|user2 |siteinfo |
|user3 |moodle_mobile_app |
|user4 |siteinfo |
When Iloginas"admin"
And Inavigateto"Server>Webservices>Managetokens"insiteadministration
# All created tokens are shown by default.
And "Firstname1Lastname1""table_row"shouldnotexist
And Ishouldsee"Siteinformation"inthe"Firstname2Lastname2""table_row"
And Ishouldsee"Moodlemobilewebservice"inthe"Firstname3Lastname3""table_row"
And Ishouldsee"Siteinformation"inthe"Firstname4Lastname4""table_row"
# Filter tokens by user (note we can select the user by the identity field here).
When Iclickon"Tokensfilter""link"
And Isetthefield"User"to"user2@example.com"
And Ipress"Showonlymatchingtokens"
Then "Firstname3Lastname3""table_row"shouldnotexist
And "Firstname4Lastname4""table_row"shouldnotexist
And Ishouldsee"Siteinformation"inthe"Firstname2Lastname2""table_row"
# Reset the filter.
And Ipress"Showalltokens"
And Ishouldsee"Siteinformation"inthe"Firstname2Lastname2""table_row"
And Ishouldsee"Moodlemobilewebservice"inthe"Firstname3Lastname3""table_row"
And Ishouldsee"Siteinformation"inthe"Firstname4Lastname4""table_row"
# Filter tokens by service.
And Iclickon"Tokensfilter""link"
And Isetthefield"Service"to"Siteinformation"
And Ipress"Showonlymatchingtokens"
And Ishouldsee"Siteinformation"inthe"Firstname2Lastname2""table_row"
And Ishouldsee"Siteinformation"inthe"Firstname4Lastname4""table_row"
And "Firstname3Lastname3""table_row"shouldnotexist
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Jun Pataleta
