* When restoring backup we clean the contentformat and defaultformat
to avoid issue with invalid value.
* Add tests for format and defaultformat so to check that we
filter out invalid formats
* Remove private $basepath variable as it was used one and not always
initialised correctly.
* A presentation role is not necessary for the img tag.
- If a non-empty alt text is provided, the presentation role will
conflict with the alt text.
- An empty alt text denotes a decorative image. The presence of a
presentation role is redundant.
* Make sure that the alt text is set to an empty string. Otherwise,
an img tag without an alt attribute will be produced which is an
invalid markup.
* When text is rendered for the action icon, set the icon as a
decorative image by setting empty alt and title attributes and adding
the aria-hidden attribute as well. Otherwise, assistive technologies
will read the action name twice.
* Axe DevTools flags an error on the recently accessed items block that
the listitem role is not allowed for <a> tags. This is because the
"listitem" role will override the native semantic meaning of the course
links. So the solution is to move the course links within a <div> tag
with the listitem role instead.
The non-empty alt text for the icons of the context header buttons like
the "Message" and the "Add contact" buttons cause redundant information
to be read twice by screen readers when the buttons are rendered with
the icon and the button name. The icons need to be set as decorative by
setting an empty alt text for them.
* The presentation role is also unnecessary.
* The alt text is needed when the button gets rendered without the
button name text.
* Decorative images should have an empty alt text and there's no need
to set a presentation role.
* Accessibility Behat tests added to cover the changes
* Decorative images should have an empty alt text and there's no need
to set a presentation role.
* Accessibility Behat tests added to cover the changes
* Decorative images should have an empty alt text and there's no need
to set a presentation role.
* Accessibility Behat tests added to cover the changes:
- block_starredcourses uses the core_course/no-courses template
- core_completion tests changes in core_course/activity instance and
core_course/editdefaultcompletion
* The zero state image is purely decorative so there's no need to
set a non-empty alt text for it.
* The presentation role is also unnecessary for decorative image with
empty alt.
* Accessibility Behat tests added to cover changes
* The zero state image is purely decorative so there's no need to
set a non-empty alt text for it.
* The presentation role is also unnecessary for decorative image with
empty alt.
* Accessibility Behat tests added to cover changes
* An image with non-empty alt text should not have a presentation role.
* An empty alt attribute is sufficient for decorative images. No need to
add a presentation role for them.
* Additionally, if there is already text that is present for an image,
there's no need to add an alt text with the same text. Such images are
decorative and should just have an empty alt text.
This is a temporary solution to the accessibility problem
"Fieldsets without legend". We will no longer use a fieldset tag
if there is no legend after MDL-80431
When accessing a datepicker that is within a dropdown menu using the tab
key, the navigation sequence was changed since the datepicker code is
not rendered within the dropdown menu element, thus the dropdown was
closed as it was understood that we were not more inside it.
Furthermore, when exiting the datepicker by pressing tab, as the
dropdown was closed, the navigation jumped to the first element on the
page, which is the link to Skip to main content.
To resolve this, a validation was added that keeps the dropdown menu
open if the datepicker is open and accessed via the tab key.
If $CFG->enable_read_only_sessions_debug was not enabled, debugging could be
enabled part-way through a request when restart_with_write_lock was
called. This meant that a diff between the initial and final session
would be made during write_close(), although the intial session state
was never captured. This generated false positives in the logs, and it
thought any variable set in the session was a change from the original
value.
This ensures that debugging is enabled before the debug flag is allowed
to change, preventing false positives.
This was always bad, but since we only used it to verify the existence
of the param, and let library code take $_REQUEST for the real
validation, it was ok. Now, since we're redirecting to self during
necessary cookie checks added by MDL-80835, we lose the real value
originally stored in $_REQUEST. This patch just fixes the param type,
setting it to raw, which is what it should have used originally. The
raw value won't be cast from a string to an int as part of the
require_param call, so the value won't be lost any more.
The original implementation of group filtering introduced in MDL-80565
assumed that the log table existed in Moodle's own database. This is not
the case of the database logstore, or any similar logstore implemetning
the database \core\log\sql_reader interface.
Furthermore this check was also applying the SQL when the user had the
`accessallgroups` capability, or when the course was not in SEPARATE
groups mode (no groupmode and/or visible groups).
Co-authored: Laurent David <laurent.david@moodle.com>
The \Traversable interface does not define the `valid` method.
Furthermore, the `valid` method actually checks that there is a _next_
value, which requires the value already be fetched and waiting. This is
not the case for all Iterators. For example the CallbackFilterIterator
does not load the initial value until it is requested.
It is completely unnecessary to do this check anyway as an invalid
Iterator will just not return any values.
This change reverts much of MDL-74641, removing the erroneous userquota
checks from zip_packer. Now, during extraction, the zip_packer will exit
with an error status when it encounters a file which is larger than its
advertised size, and NOT when it determines the size of the extracted
files have exceeded the userquota (which only applies to private files
and which was used incorrectly here). It also reverts changes to the
draftfiles_ajax.php file, restoring the original (pre-MDL-74641) error
handling.
* When creating default templates, avoid escaping the field names
contained within [[ ]] brackets;
* When managing field, ensure the field names are escaped.
Drag and drop overlay in course page was being overlapped by divider
buttons and quick edit links.
That happens because both elements have their own z-index context.
Add z-index to the drag and drop overlay elenment to fix this issue.
The MoodleSession cookie is already set this way when using Chrome, but
not for Firefox. This is a decision made in \core\session\manager. When
Firefox eventually defaults to SameSite=Lax when the SameSite attribute
is omitted from the Set-Cookie headers, the MoodleSession cookie won't
be sent. This change ensures the cookies used during LTI launches will
continue to work if/when this happens, and despite any changes which may
be made to core's sessionlib in future.
Adds the property that is required by Chrome to opt-in to its 3rd party
cookie partitioning solution, CHIPS. This specific change ensures the
'state' cookie, used in the OIDC handshake, has partitioning support.
This cookie can be partitioned unconditionally, since it's a cookie
controlled by the library and one we don't expect to be set without
partitioning elsewhere.
Adds the property that is required by Chrome to opt-in to its 3rd party
cookie partitioning solution, CHIPS. This specific change to auth_lti is
to ensure the MoodleSession Set-Cookie header resulting from
complete_user_login() calls (in auth.php) have this property set.
Adds the property that is required by Chrome to opt-in to its 3rd party
cookie partitioning solution, CHIPS. This specific change deals with the
cookie that is set when the user is not yet auth'd with the site and is
necessary to facilitate OIDC nonce retrieval and validation.
To opt a cookie in to Chrome's 3rd party cookie partitioning solution,
CHIPS, the property 'Partitioned;' needs to be set. This adds a helper
class supporting this, for a given cookie(s).
Note also, PHP's native
cookie APIs (setcookie, etc) don't support this cookie property yet -
(https://github.com/php/php-src/issues/12646).
Since this class is intended to allow existing Set-Cookie headers to be
modified before being sent (e.g. allowing clients to set a property on a
cookie set elsewhere in code), it deals with the headers directly anyway
but it means that new cookies must also use this helper to opt-in,
instead of relying on setcookie(). E.g. where the intent is to add
partitioning support to a new cookie, that cookie must first be set
(setcookie) and then it may opt-in to partitioning via this helper;
partitioning support cannot be achieved directly through setcookie and
friends yet.
This adds a few changes to the old test_context_not_used test:
- Move it to become an advanced_test, because its mission
is to verify that the assertEventContextNotUsed() assertion
works as expected.
- For consistency, also move the fixtures to own phpunit fixtures.
- Add proper coverage tags, to verify that the assertion is being
covered.
- Add a data provider to provide all the current cases and ease
any future case that may be needed in the future. One by one
because previously there was code never executed with the
warning expectation causing the test to stop.
- Run them in isolation, while this is not strictly required, it's
including external fixtures and, we'll need that isolation soon
(for changes coming when moving the test to PHPUnit 9.6 in MDL-81266).
When the user does not have a profile photo and initials are displayed,
there is no alt text for the initials which causes accessibility
issues, so with this change I added the user's full name to the title
and aria-label attributes and I set a role='img' when the element span
is rendered.
As part of implementing this, there is a new more efficient method
get_version_information_for_questions_in_attempt to see if any
questions in a quiz need to be updated. The regrade code was updated
to take advantage of this.
Co-authored-by: Khoa Nguyen Dang <khoanguyend7@nashtechglobal.com>
The column type determines the available aggregation options present
for the column. Custom fields of type "select" should be considered as
text fields for this purpose, because their stored value represents
the index to their available options, rather than having any distinct
meaning of it's own for display.
If the list of site admins was defined in the config.php, the UI to
select / modify admins never worked. But there was no indication of the
reason which was unexpected and confusing.
* forcing new generated sets of wildcard values into the question edit form so they can be saved.
* added Behat test to check new generated values are used.
Apart from environmental stuff, the moodle_minimum_php_version_is_met()
function is executed by all installs (web, cli, database, ...)
and halts any installation (early) not meeting the PHP constraints.
Before this change the query used an index scan to perform the
export, this could be really expensive on the database.
After the change the the query will use far more efficient joins,
on a large MySQL instance this can take minutes off of the query
time.
In additon the query was returning two columns that are not used
in the function. The layout column was causing the UNION to fail
on Oracle as it cannot compare text columns, removing these
columns from the returned data should further increase
performance further.
The existing query was inefficient using an index scan and
index merges. This change splits the query up into many individually
much more efficient queries.
This should significantly improve performance on large sites.
The UNION caused the query to be run in a way that is very
inefficient on MySQL, separating the queries causes each of them to
run in a much more efficient form.
Any duplicated will be filtered out on the PHP side instead of in
the database. On large Moodle sites this is preferable as the
extract is likely to be performed on a server dedicated to running
the Moodle cron and so there will be less of effect on resources
that are used to serve end users.
The UNION caused the query to be run in a way that is very
inefficient on MySQL, separating the queries causes each of them to
run in a much more efficient form.
Any duplicated will be filtered out on the PHP side instead of in
the database. On large Moodle sites this is preferable as the
extract is likely to be performed on a server dedicated to running
the Moodle cron and so there will be less of effect on resources
that are used to serve end users.
This change removes LEFT JOINS to improve the performance
when run against Moodle sites with very large data sets.
It will also cause some additional contexts to be returned:
1. When a user has modified a user_evidence_competency
without modifying its associated user_evidence record
Previously, groups with GROUP_VISIBILITY_NONE could be selected for an
availability condition, but as the group API did not return a user's own
groups with this visibility, the condition's is_available check always
failed.
This change uses the new $includehidden parameter to get all of a user's
groups from groups_get_user_groups when evaluating the condition, so
these groups work as expected. Due to the enforced hiding of conditions
containing these groups, they will never be seen by the user.
This allows groups_get_user_groups to return groups with
GROUP_VISIBILITY_NONE that the specified user belongs to. The parameter
is false by default, and should be use with care as membership of these
groups should not be revealed to the user.
Invert the previous logic, which maintained a "banned" list of
fields, to instead only try to filter on those fields that are
explicitly defined to support that.
To minimise performance issues when displaying courses in Summary view,
we need to conditionally show the "All" paging option. This commit
will prevent the "All" paging option to be shown if the following
conditions are met:
- The display mode is "summary"; and
- There are more than 100 courses to be shown.
The datepicker is not keyboard accessible when in a modal.
As we are using yui datepicker and the change is not simple, I added aria-hidden and tabindex properties to avoid screenreaders.
Move the `get_name` method to the base report class, so that it can be
implemented by all report types. Provide a default implementation for
system reports based on the name of the class.
Use the name of the report as the caption. Target via Behat selectors.
* added calculated question for multiplication to tests/helper.php.
* added test_grading_of_negative_responses() test to question/type/calculated/tests/question_test.php
The only tests that have COMPLETE* support for the debugging sink
are the advanced_testcase and the database_driver ones (store and
report). So we must ensure that the rest of tests don't use the
debugging sink at all.
Right now we are using it for storing, but later there is not
reporting, so any debugging happening within non advanced tests
is not detected.
This commit just ensures that we stop making that storing for
non advanced/database_driver tests. Nothing more, nothing less.
* Note that we have had to add a few missing bits to the
database_driver testcase because it was not 100% complete. Now
it behaves 100% the same than the advanced_testcase one regarding
the debugging sink.
This patch fixes the following error:
PHP Deprecated: explode(): Passing null to parameter #2 ($string)
of type string is deprecated in lib/upgradelib.php on line 1299
Most types of form field will now include aria-required="true" if the
field is marked as required. This causes assistive technology to inform
users that the field is required.
Before this change, in some cases (e.g. screen reader users tabbing
through fields) users were not informed that a field is required.
Browsers are phasing out 3rd party cookies. Those which can be set are
partitioned to the top level embedding site, so piggybacking is
prevented. This will break the account linking process. This fix swaps
the piggyback for a login round trip, as originally intended, which
resolves the issue.
Add missing entry for custom $CFG->localrequestdir locations in the
preconfigured array for $CFG->xsendfilealiases in config-dist.php.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
We should not rely solely on the provided size information.
Therefore the system needs to recalculate the size of each file in the zip file,
whether small or big.
The changes also cover the filemanager.js file if an error appears.
Prior to this change, assertions of "should exist" and "should not
exist" in table content, where the step provided only a single column,
could give false positives and pass (when they shouldn't).
This commit is upgrading the joubel/core library to the one
tagged as 1.26 (which also increases the coreApi minorVersion).
That's why the previous commit was duplicating the existing
h5plib_v124.
This tag is also pointing to the same tag that the mod_hvp is using
too: https://github.com/h5p/moodle-mod_hvp/releases/tag/1.26.1
- Added a new method called get_messages_by_component()
This method will help to retreive the redirected messages
of specific component only
- Added a new method called get_messages_by_component_and_type()
This method will help to retreive the redirected messages
of specific component and type only
It should always be the case that a question category and its parent
category belong to the same context. However, over the years, we have
observed that this becomes violated by some data in some Moodle sites.
(The suspicion is that this is a backup/restor bug of some sort which
has never been tracked down.)
Since this happens, we should make the code robust to it, which is what
this change does. It ensure that when loading child categories, we only
consider ones in the same context.
The new API replaces identical behaviour in existing calling code, but
allows for profile field types to override/separate the logic used to
show the field and determine whether it's empty
* Update \behat_accessibility::run_axe_validation_for_tags()'s
PHPDoc block to reflect the current version.
* Fix incorrect copyright tag for the library
Stylelint 15 deprecates a heap of rules which will be removed in 16.
We can't use 16 because it's ESM and requires a big restructure of our
build process.
We can temporarily silence the deprecations and then when we update our
build tooling we can make use of @stylistic/stylelint-plugin to add the
rules back.
This commit:
- bumps stylelint and related dependencies accordingly
- fixes issues identified with newer sniffs
- temporarily silences deprecation warnings
There are two phases of a build: Building, and then Outputting.
We were previously listening on the final event for the build phase, but
we should be listening to the final event of the output phase.
grunt-jsdoc is abandoned and only works with an older version of jsdoc.
This is a very simple wrapper around jsdoc itself.
This commit also includes a fix for broken docs.
This commit is a backport of MDL-74301 which should have been backported
at the time.
All rules included in "eslint:recommended" set containing matching
values are removed from the list. Recommended set is defined in
node_modules/eslint/conf/eslint-recommended.js
This commit is a backport of MDL-74301 which should have been backported
at the time.
To identify deprecated rules, temporary install sarbbottam/eslint-find-rules, then run
`npx eslint-find-rules -d .eslintrc`, it will return a list of rules
defined in .eslintrc that were deprecated.
This commit is a backport of MDL-74301 which should have been backported
at the time.
Adding 'plugin:promise/recommended' results in 5k+ jsdoc warnings most
of which includes missing return and naming preferences (e.g. 'returns'
over 'return' or 'string' over 'String'). This needs to be audited and
addressed seprately.
This commit is a backport of MDL-74301 which should have been backported
at the time.
This enables recommended rules and brings back related rules. Without
reducing level to warn for some of them, this gives numerous (about 88) issues mainly
catch-or-return. Recommended set is defined in
node_modules/eslint-plugin-promise/index.js
This commit is a backport of MDL-74301 which should have been backported
at the time.
This patch includes changes:
* Upgrade "@babel/core" to latest point release
* Remove plugins already included into current "@babel/preset-env":
- "@babel/plugin-proposal-class-properties"
- "@babel/plugin-proposal-json-strings"
- "@babel/plugin-syntax-dynamic-import"
- "@babel/plugin-syntax-import-meta"
* Upgrade "eslint" to latest version
* Replace "eslint-plugin-babel" (depreacted in 2019) with "@babel/eslint-parser" and "@babel/eslint-plugin"
* Changing the name of an activity should change the access restriction messages.
* Introduction of a marker as data-cm-name-for to be able to change any
occurence of a course module name in the page when this one changes.
The UX team proposal is to display two different notifications, to avoid confusion:
1. When the user is in preview mode:
Message: 'You are in preview mode.'
Notification Type: Information (blue), with the information permanently visible and not dismissible,
ensuring continuous awareness.
2. When activity tracking is disabled:
Message: 'Attempt tracking is not enabled for this activity. You can enable it in Settings.'
Notification Type: Warning (yellow) with a close icon for dismissal if tracking is intentionally disabled.
These notifications should only be visible to teachers and content creators, not students.
If a teacher also has a student role, this information should not appear when logged in as a student.
That's why the pre-existing is_tracking_enabled() function has been updated to only check if tracking
is on for an activity, without considering user capabilities (that's why the $user parameter has been
removed from this function too). Besides, a new function called can_submit() has been created to find
out if users are allowed to submit an activity.
The aria-labelledby attribute for the action menu link's <a> tag is
unnecessary as the <span> tag containing the action menu link's text
is already within the <a> tag and already serves as the link's label by
default.
For sites that rely on environment variables to set configuration,
ensure they are also present when executing task process.
Co-authored-by: Olivier Wenger <olivier.wenger@liip.ch>
Occurs when a plugin in incorrectly removed (via file system only,
rather than following proper uninstall process), and also contains
it's own sub-plugins.
In the above scenario, the update would request and receive back a
response for a plugin whose name was comprised of an empty string,
subsequently causing errors when processing said response.
If an existing session is active at the time of launch, that user id
appears in the user_updated log. This log should show the launching user
updating their own record, not someone else. To resolve this, auth the
user before making any record updates so that the event gets the correct
user id.
If an existing session is active at the time of launch, that user id
appears in the user_created log, incorrectly (it should be system/0).
This patch makes sure that any existing session is terminated before
creation of a new user during launches.
Instead of manually including composer's auto-loading
stuff and then run PHPUnit\TextUI\Command::main(), now
we are using the PHPUnit "binary".
That way behaviour should be 100% the same than
running vendor/bin/phpunit (recommended instead of util.php --run).
Also, note that, for other util.php commands, we still need
composer's auto-loading to happen, so we have moved it immediately
after the --run command.
Finally, a few more file_exists() conditions have been added
to ensure that PHPUnit is properly installed. Redundant but...
In the course external create/update methods, we must ensure that
the correct form element names are used. Specifically because the
textarea field type uses non-standard name.
This ensures compliance with the following WCAG criterion:
"H71: Providing a description for groups of form controls using fieldset
and legend elements".
This is consistent with the API used for other form element types,
in order to provide screen-readable text in the fieldset legend while
not necessarily wanting to display said label.
The new class property is automatically exported to the template via
the `templatable_form_element` trait.
The button at the top of the page is hidden a few seconds by the
un-enrolment confirmation dialogue. Instead of waiting a few seconds
or reloading the page, the button at the bottom is clicked instead.
It's better for the task to fail gracefully rather than silently
swallow any errors, so that the task API can schedule re-tries.
If the endpoint is temporarily down and/or misbehaving, this will
allow admins to have clearer failures.
Also correctly extract the original file extension.
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
Co-authored-by: Mazitov Artem <maz@2i.tusur.ru>
Implement grade validation for numeric grade overrides by setting
the type attribute for the grade input to number and setting a min
and max attribute values. Submitting the form with invalid values
will trigger built-in validation of the inputs.
* First parameter should be the field name that the format property
points to.
* No need for VALUE_REQUIRED since it's the default for the $required
parameter.
* Third parameter is the default value and not a description string.
* 'Content format' is passed to the $default parameter which is
incorrect.
* Also removed passing of VALUE_REQUIRED because it's already the
default value for the $required parameter.
- If a text format property defines the 'optional' attribute as true,
then the $required parameter that
is passed to \core\external\exporter::get_format_structure() should be
VALUE_OPTIONAL.
- If a default value is provided for a format property,
\core\external\exporter::get_format_structure() should pass the default
value to the external_format_value constructor.
- Added validation for external_format_value for the default parameter.
Debugging will be shown in case an invalid default value is passed.
- Amended unit tests for better coverage.
Since 4be39296 the root HTML element contains the iso6391 language
code of the current language pack. However various JS modules need
access to the original Moodle language pack code, so expose that
in the `M.cfg` structure.
When xdebug is set to break on exceptions it stops on all exceptions,
including _caught_ exceptions.
As a result we regularly hit this uniqid helper issue if debugging with
xdebug.
The solution is to check if the helper exists using the hasHelper method
instead of catching an exception when it does not exist.
This random error was caused because when reviewing an attempt,
the first pageid to display is calculated based on the timeseen
field in the lesson_attempst table.
In Oracle, sometimes it returned the second pageid (because both
had the same timeseen), so which led to these random failures.
Apart from adding 1-second wait when the attempt is created (to
guarantee the first page is returned properly), a few more
improvements have been also made, like removing manual steps for
setting the lesson activity or adding some extra checks to
confirm the information displayed is expected.
This works around a problem in SQL Server, which caused it to throw a
DML exception if the same column is duplicated in `ORDER BY` clause.
Backported from MDL-79270.
Tabs use the value of the data-target attribute of the active tab link
to find and display the appropriate tab content. In some situations the
elementid value that is used to generate the data-target may contain
certain characters that have a special meaning in the selector context
(e.g. ".", ",", "#") and because of that the target element cannot be
correctly identified. To fix this problem the elementid value needs to
be escaped prior to generating the data-target attribute.
The addition of text format on the output of the answer responses broke
the matching later when comparing the valid answer against the sent
response, as the answer response was not correspondingly formatted.
I decided to use the un formatted answer response as the key, to keep it
as close as identical to prior behaviour
* Create SRLogger that extends from the default Logger class. This
logger outputs the feedback to a sr-only ARIA live region to allow
screen readers to announce the feedback from mutations.
* Update the course editor to use the SRLogger.
* Create a mechanism for logging mutation feedbacks.
* The mutation feedbacks are displayed as a toast by default.
* Apply this logging mechanism on the course homepage to provide
feedback for the results of actions performed on course modules and
course sections.
No need to pass a value for the header parameter in
print_grade_page_head() because:
1. We don't really render a header on that page.
2. It will allow the use of the editingmode_title lang string for the
grader report so the page title can reflect the display mode of the
grader report, especially when editing mode is turned on.
There are checks to urls that attempt to limit recurrsion when
parse_file is called. This is problematic for css import urls that
can call an indefinite amount of nested import urls. An import limit
has been introduced to address this. Fragments have also been removed.
get_gradable_users now has extra param to retrieve only active
users. Grade reports hase their own grade_report::get_gradable_users
since they decided whether to inlcude active users from report preferences
We should verify that schedule and audience users are active accounts,
and that we don't try to send report schedules either from/to them if
they aren't.
If a quiz activity has negative grading enabled for incorrect answers,
students can achieve overall negative grades in this quiz. If at least
one student achieves a negative grade, the chart at the bottom of the
quiz 'Results' tab will fail to display. This patch add a search for
negative results, removes them from the band below 0 and adds them to
the 0 band. This will make the chart render correctly again.
Co-authored-by: Susana Leitão <sleitao@uporto.pt>
Co-authored-by: Matthias Opitz <m.opitz@ucl.ac.uk>
Removed whitespace at end of lines
* Add support for grade report plugins that define an
`editingmode_title` lang string. If the grade report plugin defines this
language string, it will be used for the page title's unique
identifying information when editing mode is turned on.
* Make the headings for the single view easier to read and understand
* When on editing mode, update the heading for the grade and user
screens to convey to the user that the page is currently on editing
mode. This is helpful for screen reader users as they can immediately
know the current display mode of the single view report page when this
gets read via the page title.
Besides adding the is_guest() check in the check_password_policy() condition,
I also changed the temporary user ID because the temporary user and
the guest must not share the same ID. Otherwise, it will raise an error.
Co-authored-by: Andrei Bautu <abautu@gmail.com>
In certain scenarios, a user row can be the last row in the grader
report table. The current logic does not take this into account and
always pins the last row regardless of the type.
This change ensures that the last row is not a user row before executing
the logic responsible for pinning this row.
In pages which result in the theme_boost/embedded template being used,
blocks may be rendered and the resulting HTML ignored, leaving any JS
they may have added to run without a corresponding DOM element to act
on. This fixes that case for comments, making sure we exit gracefully
during the init if the element can't be found.
The new activity card design, planned for Moodle 4.3, reintroduces
the resource MIME icon that was removed in Moodle 4.0.
This is a backport of MDL-78285.
The placeholders for the hints are properly replaced by this change and any formulas present are replaced.
The validation is performed by the existing function qtype_calculated_find_formula_errors(),
which is called by the calculate() function,
which in turn is called by the variable substituter "vs" using the replace_expressions_in_text() function.
vs -> replace_expressions_in_text() -> calculate() -> qtype_calculated_find_formula_errors().
Previously, $CFG->sessionlockdebug was reporting long session locks for
all requests, even if they defined READ_ONLY_SESSION. This change
ensures that only sessions with locks are reported.
This resolves accessibility issues where Moodle language pack codes
didn't always map to correspondingly named iso6391 codes.
Where this value is defined in the language configuration, it will
now be used.
The PSR-4 specification does not preclude a single namespaces from
having multiple sources. This is the case for several PSR standards
including standard packagist packages distributed by PSR:
- PSR-7 - HTTP Message Interfaces: \Psr\Http\Message
- PSR-17 - HTTP Factories: \Psr\Http\Message
- PSR-15 - http-server-handler: \Psr\Http\Server
- PSR-15 - http-server-middleware: \Psr\Http\Server
With the h2 heading not being rendered on the zero-state pages of the
Single view report, we need to increase the heading level of the
"Search for a user to view all their grades" and
"Select a grade item above" headings from h3 to h2.
In order to prevent rendering an empty heading <h2>, we need to make
sure that the $heading parameter passed to print_grade_page_head() is
not an empty string.
PHP 8.3 deprecates the use of `get_class()` without arguments. The
following alternatives should instead be used as appropriate:
- `get_class($this)` - The equivalent for instance methods
- `self::class` - The current class
Note that the behaviour of `get_class()` in a static method was
equivalent to `self::class` and _not_ `static::class`. That is to say
that the previous behaviour did not respect late static binding.
- Fix the arrow placement for bootstrap popovers in RTL mode.
- Remove unneded css parser workaround for popover arrows.
After MDL-67386 was integrated those changes were not needed anymore.
The metadata information was not restored when an H5P file was uploaded
(it was saved when it was edited directly from the Content bank, but the
original information in the H5P file was lost).
This patch includes all the metadata fields and its values in the original
H5P file to preserve this information too.
The Course backup area was not displaying the course backups when
accessing to the Restore page for an activity because the context
was not updated properly for this area.
* Feature: We can use a minimum grade different than zero
- Use correct minimum input values.
- Use data generators for faster test execution.
* Scenario: Disable category overriding
- Use the field name to ensure that the correct field is being tested.
* <h6> tags cannot be in table cells. Replace them with <div> tags.
* Group the ascending/descending sort menu items to better indicate
where they belong to.
* Make sure that we respect the fullnamedisplay and alternativefullnameformat
parameters to decide on the initials for a given user
* Add further tests
Co-authored-by: Tatsunori Uchino <tats.u@live.jp>
When bulk-updating course module visibility, set_coursemodule_visibily
was triggering a partial cache purge and rebuild for each course module.
This potentially led to 2 cache sets each requiring a lock to be
acquired and released per course module, plus any other cache updates
for other changes to the course in the same request.
This adds a new $rebuildcache paramter to the
set_coursemodule_visibilty, which is true by default to retain the
existing behaviour. If set false, it will skip doing the partial purge
and rebuild for that course module, and it is up to the calling code to
ensure the cache is updated as requried.
To assist with this, there is a new
course_modinfo::purge_course_module_cache_multiple() method, which
allows multiple course modules to be purged from the cache in a single
cache set.
Based on the feedback given by the UX team:
- The text has been increased to display more (50 vs 55).
- A tooltip has been added to display the full course name on hover,
following the accessibility guidelines.
* When reloading a section, we also forced the reloading of
course modules, resulting in doing the same work twice (once for
the section reload and once for the course module reload)
* Debounce the reloading process so prevent unwanted duplicate
reloads
The UX team confirmed a badge shouldn't be awarded when the criteria
use a passing grade and the user gets a failing grade.
So the COMPLETION_COMPLETE_FAIL status won't be considered
completed for activities with completion that require a passing
grade.
This addresses a random failure with the combobox search results where
the debounce causes the results to be shown, and then the same search
result is returned again, re-rendered, and replaced after Behat has
moved on.
Partly cherry-picked from MDL-78779.
The "flag" field isn't a boolean state (1/0), it's a cumulative
count of how many times the tag has been flagged by users. This
should be accounted for to ensure that column aggregation and
filtering performs correctly.
This changes the cache purge code to select IDs for deletion
first, rather than using a subquery in a DELETE, since this
risks causing a deadlock in MySQL.
* The arrow characters in link_arrow_right() and link_arrow_left()
functions get announced by screen readers. This causes confusion
and is unnecessary.
We no longer need special handling of this field type, because it
now returns language strings rather than HTML markup for a checkbox
element on display.
The correct way to get an ad-hoc task to run as the main admin account
is to leave userid as null.
Setting it causes fatal errors if the main admin account is set to
auth: nologin.
The UX team confirmed that an activity shouldn't be considered completed
when the criteria use a passing grade and the user gets a failing grade.
So the COMPLETION_COMPLETE_FAIL status won't be considered
completed from now on.
When we have a pagination on grader page and search for user
that is not on first page, saving was throwing an error
since page didn't have enough data
During the bootstrap of PHPUnit we ensure that the database has been
reset to its initial state.
We do this by checking the internally-stored DB write count between
runs. If the count is not yet set (null), or it has been increased, we
force a reset.
When running an isolated test the test runner resets the database, it
then sets up a new isolated test environment by writing a new PHPUnit
test case and passing it to a new PHP Process using standard in. As part
of this, the bootstrap is run for that process.
Because we are in a new process, the db write count is fresh and not yet
set. This has been leading to an additional db reset before the isolated
test.
To handle this we want to _not_ perform a reset during the
initialisation for isolated runs. We know that the DB is in a fresh
state before we start the run.
To support this we need to know whether the test is an isolated test
during the bootstrap, which means we cannot use the previous approach to
calculating this.
Instead we look at the PHP_SELF value. PHP sets this to "Standard input
code" when run from stdin, instead of running a file.
There should not be any other legitimate reason to run a PHPUnit
bootstrap via this stdin approach.
Unfortunately this approach is a little bit risky as it depends on the
presence of a specific string, however this string has been in place
since 2016, and there is no legitimate way of calculating this.
I did consider looking at whether the called script included `/vendor/`
and `/phpunit`, but this is also likely a risky approach if someone
calls PHPUnit in an unexpected way.
This approach is itself unit tested so any change to PHP's stdin string
before we deprecate this approach entirely in 12 months time will be
caught.
This hack was introduced to work around a bug in MySQL 5.6.14 and
MariaDB at the time.
https://bugs.mysql.com/bug.php?id=69882
It was addressed a few months later in 5.6.16, and 5.7.4.
MariaDB merged version 5.6.16 of MySQL's InnoDB engine in MariaDB
10.0.11 and got hte patch from there.
Moodle has required MySQL 8.0, and MariaDB 10.6.7 since Moodle 4.2 and
it is therefore safe to remove these hacks.
* In matching types of questions, sometimes the user does not answer all
questions. We now display "None" in the result report. The answer can
be either correct or incorrect.
* Fix "unknown" answer constant name.
Group visibility was not taken into account when
generating SQL for getting enrolled users restricted
to a list of groups. This may have allowed users to
infer membership of groups they were not allowed to
see members of.
In some places we prevented cache poisoning, in others we did not. We
also did not place any restriction on the minimum value for a revision.
This change introduces a new set of functions for configonly endpoints
which validates the revision numbers passed in. If the revision is
either too old, or too new, it is rejected and the file content is not
cached. The content is still served, but caching headers are not sent,
and any local storage caching is prevented.
The current time is used as the maximum version, with 60 seconds added
to allow for any clock skew between cluster nodes. Previously some
locations used one hour, but there should never be such a large clock
skew on a correctly configured system.
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
* We need to set a different meetingID when duplicating or importing a
BigBlueButtonBN activity to prevent unwanted sharing of recordings between
meetings.
With the new flow, users can go to the login page from the confirmed page,
and if the user successfully logs in, the user will be directed to the confirmed page.
To avoid that, the confirmed page can only be seen by users who are not logged in.
The Moodle XML export of multiple choice calculated questions was
missing proper escaping of answers. Also the indentation of the
generated xml was off.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
1.9 backups used current plugin version as oldversion.
But quiz uses some hardcoded version numbers when processing
restore which is lower then current plugin number, so some quiz
logic was ignored.
See define_structure(), process_quiz_question_instance() within
restore_quiz_stepslib.php
Within the forum_get_discussion_neighbours() function of forum
when the discussions timemodified (last post) is the same, there
is a bug that does strange things when calculating the prev and
next discussions.
Note that, in real life, this is really hard to achieve, but in tests,
when multiple discussions and post can be created by generators in the
same second (specially when the test machine is quick), chances of
facing that problem are higher.
By adding 1 second wait, we ensure that the discussions won't have
the same timemodified (last post) and workaround the problem.
No mater of that, the problem deserves an issue to be created
so we guarantee from code that it also works ok when the same
timemodified (last post) situation happens.
Note that normally this doesn't matter much, but there are situations
when we want the discussion list ordering fully consistent /
deterministic.
Specifically, when discussions (or forum posts )are created in the
same second, or when the discussion titles are repeated, or 2
discussions have the same number of votes... (any criteria in general),
in the context of testing, we don't want the order
to be non-consistent, so we need to provide an extra sorting
criterion to make it fully deterministic.
So, in this case, we are adding a sort by discussion.id <<DIRECTION>>
that is an unique value, primary key... so cheap to calculate and,
that way, when the 1st sorting column has repeated values, the id
will decide.
The selection gets lost while opening the modal dialogue to update an
embedded media. Caching the current selection allows us to update the
previously selected node instead of updating the first embedded media.
Signed-off-by: Gregor Eichelberger <gregor.eichelberger@tuwien.ac.at>
The short name of the cc licenses are trailed with a suffix containing
the version number (current 3.0 and 4.0). The old cc* licenses become
the new cc-*-3.0 licenses and are disables, because the new cc*-4.0
licenses are the current ones.
This is a backport of MDL-43195.
Associate column headings of user report tables to the user ID that the
report is being generated for. This will prevent duplicate IDs for the
report columns when multiple user report tables are displayed on the
User report page.
* The table cells mentioned in the `headers` attribute of other table
cells need to be using the table header <th> tag. Otherwise, HTML
validation fails.
when `log_out` is called from `\core\oauth2\client` it will delete the refresh token,
what it actually needs to use it to get a new access token
actually logging out is not needed here, the only thing we need to make sure is,
the invalid access token is removed from the session
that is done by storing `null`
- Switch use PARAM_TEXT instead of PARAM_URL for resource URL
- Added noreferrer to the Go to MoodleNet drafts button, to avoid the risks associated with opening in
_blank without removing access to the referrer and opener
Currently, Moodle uses fetchOBject() from ADOdb to get the table columns.
Sadly, the current ADOdb, especially the fetchObject() function, still creates dynamic properties.
Altering the fetchObject() with fetchRow() to avoid the dynamic properties deprecation error on PHP 8.2
Currently, Moodle uses fetchOBject() from ADOdb to get the table columns.
Sadly, the current ADOdb, especially the fetchObject() function, still creates dynamic properties.
Altering the fetchObject() with fetchRow() to avoid the dynamic properties deprecation error on PHP 8.2
Previously, quiz statistics processing happened on a scheduled task.
This task looked for all quizzes with completed attempts, then
determined if those quizzes had a statistics calculation that's newer
than the most recent attempt, then ran the statistics calculation if
needed. It was hard coded to stop processing after 1 hour.
The queries involved in determining which quizzes needed processing
weren't terribly efficient, and combined with the 1 hour limit this made
the statistics unusable on large sites, where they are the most useful.
This converts the scheduled task to an ad-hoc task, and uses an event
observer for mod_quiz\event\attempt_submitted to queue a task when
it is needed. This removes the need for a query to work out what needs
processing, and allows the task processing to be scaled up as needed.
We encountered errors on the plugin check page while upgrading from version 3.11 to 4.3.
The plugins marked with the status "To be deleted" have a null value for the $component property,
which results in the following error message:
"Deprecated: strpos(): Passing null to parameter #1 ($haystack) of type string is
deprecated in lib/classes/component.php on line 1056."
To avoid this error, we need to set an empty string as the default value for the $component property.
This new parameter / property will decide if we want to reduce
the run data before processing it:
- By default it will be disabled in table mode.
- By default it will be enabled in graph mode.
- The defaults can be changed by adding reducedata=[0|1] in the URLs
- Once data reduction is enabled, it stays enabled while
navigating within the xhprof reports.
This covers the 2 new functions with unit tests:
- xhprof_topo_sort()
- reduce_run_data()
Note that the example graph used in the provider is the
one shown in the issue to explain the reduction procedure.
Here we are reducing the xhprof runs data by removing the
__Mustache==>__Mustache calls and all the orphaned data.
To save N iterations what we are doing is:
0. The information is "topologically" sorted, so we ensure that
all the parents in the data are processed before the children.
(this will help a lot when cleaning orphaned data, see below).
1. First pass, all the candidate (by regexp) calls are removed
from the run data.
2. Second pass, all the orphaned information (calls that have
ended losing his parent) are also removed, so data is consistent.
Note that, normally we would need N passes to remove all the
orphaned data (because each pass creates new orphan candidates),
but, as far as we have ensured that the information is topologically
sorted (see point 0 above), all this can be done in one unique pass.
TODO:
- Add unit tests.
- Enable some system to be able to decide which utilities we
want to get the data reduced and which ones will continue
using the complete data. Right now the reduction is being
applied to all the utilities (both table and graph views).
- Document the change and, if implemented, the way to select
between complete/reduced data.
- Consider adding some caching to speed-up the reduction process
(some TODOs have been left in the code pointing to the critical
points).
* Use a unique ID the date nodes in the HTML tree to be able to find the current node.
* Look for nodes in the same tree "leaf" and work on conflicts in this single leaf/branch.
* Use the appropriate gradebook plugin string for the page title.
* Improve gradebook page titles set by grade_get_plugin_info() by
listing the unique identifying information for the page first
(e.g. the gradebook plugin name), followed by broader information
(Grades + Course name), separating these bits of information using the
pipe `|` character.
For more information, check out
https://www.w3.org/WAI/WCAG21/Techniques/general/G88
If a quiz had a long job to calculate statstics running, this would
cause pages that may also attempt a recalculation (the statistics report
page or question bank) to load very slowly, and possibly result in a
database deadlock.
This change will firstly prevent the question bank page performing
analysis calculations at all, since these are not required for this
page, which will speed up loading and prevent deadlocks on this page.
Secondly, this adds a lock to the recalcuation process so that it cannot
run twice concurrently. This will present the user with a message to
indicate that it is waiting for a running calculation until it is
complete, and eventually it will timeout with a message and debugging.
The external test file URL concerns itself only with HTTP_USER_AGENT
matching, not sending response headers, which can differ according to
HTTP protocol in use by the endpoint (1.1 vs 2).
Given the returned response code itself is irrelevant to the testcase,
there's not much benefit to asserting it and risking random failures.
We were experiencing problems caused by old stats analysis in our
database, computed when there was a bug, causing fatal errors
when they were displayed. Therefore, this code needs to be robust to that.
The above syntax is defined as supported by the class, for example the
format '5/10' means:
"At every 10th <unit> from 5 through <max>."
It is analogous to '5-<max>/10'.
This was causing the adhoc task to fail with a nopermission error, when
a user was unenrolled from a course before their assignment submission
could be converted into a pdf.
Comments from MDL-56810 indicate the correct approach is to not attempt
a conversion in those cases.
Instead of doing an exact checking of the page title in
\behat_hooks::before_step(), do a more lenient check by checking that
the page title contains the acceptance test's site name.
* Use the page title separator constant when displaying the page title
during upgrade and installation.
* No need to display the site name during install when because it hasn't
been set at this point.
* Page titles should display the most unique information first. For
admin pages it would be useful to display the information that
is unique to the page first before the broader categories that the
page belongs to.
* Also use the new page title separator constant.
MDL-78806 core: Remove redundant site name on page titles
* With the site name now being appended to the page titles, there
is no need to manually append the page titles.
* When toggling bulk editing, updating the page's title can help users,
especially screen reader users, to determine the current editing state
of the course homepage.
* Create 'editing' versions of the coursetitle and coursesectiontitle
lang strings used for the page title of the course homepage and use
these when editing mode is turned on to provide immediate information
to users, especially screen reader users, the current state of the
course homepage.
Problem:
In Moodle 4.2, an unintended change in behaviour occurred when locking
and unlocking grade items. Previously, in Moodle 4.1, locking a grade
item and then subsequently unlocking the grade item left specific grades
for students still locked if they were being locked individually.
However, in Moodle 4.2, the unintended change caused the individual
grades to be unlocked along with the grade item.
Solution:
To address this issue, I modified the set_locked calls in action.php to
pass the cascade parameter as false. Now, when locking or unlocking a
grade item, only the grade item itself is affected, leaving individual
grades unchanged. I have also modified the locking behaviour in the
grade_category class. So, similarly, when locking or unlocking a grade
category, the change applies only to the grade items and grade
categories within the category, without expanding to individual grades.
Explanation:
By making this change, we ensure that the locking mechanism behaves as
expected, providing a consistent and desirable user experience. Locking
and unlocking a grade item or category will no longer inadvertently
affect individual grades, preserving their locked status and avoiding
loss of data when unlocking a grade item associated with the total mark
of a category.
A lock/unlock option is added to the context menu of the grade
categories again. But it is only used as an auxiliary tool to lock or
unlock all the grade items within the category.
Problem:
The grade_categories table lacks a 'locked' field, making it technically
impossible to lock grade categories. In Moodle 4.1, the "category total"
column could be locked (from the edit category form), but locking grade
categories directly was not supported.
Solution:
In response to the issue, the lock/unlock feature for grade categories
has been removed. Since the grade_categories table does not have the
necessary 'locked' field, attempting to implement category locking would
be infeasible. As a result, the feature has been removed from the
interface to avoid confusion.
Explanation:
By removing the lock/unlock feature for grade categories, we ensure that
users no longer encounter non-functional options and prevent any
potential misunderstandings about the locking behaviour. Additionally,
this change aligns the user interface with the underlying database
schema and eliminates any misleading functionality.
Problem:
When the aggregation method is set to "natural," grade items' weights
and the maximum grade of the grade category's item get recalculated,
causing the needsupdate flag to be set to true. As a result, the locking
process was skipped, leading to unexpected behaviour.
Solution:
To address the issue, the set_locked method has been modified. Instead
of skipping the locking process, the method now schedules the locking of
grade items to occur slightly in the past, specifically one second in
the past. This ensures that the grade item will be automatically locked
after the recalculations are completed.
Explanation:
By making this adjustment, we ensure that the locking process is not
skipped during natural aggregation, maintaining consistent behaviour and
preventing any unintended consequences related to grade item locking.
Problem:
Previously, when users pressed the Enter key while interacting with the
grader report, instead of submitting the form, a table column was
inadvertently expanded or collapsed. This behaviour occurred due to the
expand/collapse buttons lacking an explicit 'type' attribute within the
form. As a result, these buttons were mistakenly treated as submit
buttons.
Solution:
In this commit, we have addressed the issue by ensuring that the
expand/collapse buttons now have the correct 'type' attribute set,
explicitly identifying them as non-submit buttons. This adjustment
prevents them from interfering with form submission unintentionally.
Explanation:
By making this fix, we improve the user experience in the grade book,
allowing users to navigate and interact with the interface more
smoothly. Pressing the Enter key will now trigger the form submission as
expected, rather than inadvertently activating expand/collapse actions.
The `contextid` argument of the `tiny_equation_filter` AJAX call is
hardcoded to 1, which causes trouble if the context table does not start
with id 1.
Signed-off-by: Gregor Eichelberger <gregor.eichelberger@tuwien.ac.at>
* For some reason, the `data-trigger="hover focus"` does not work when
rendering the popover via HTML. It just defaults to click, so the
feedback popover for qtype_multianswer questions does not show on hover.
Configuring the feedback popover's trigger via JS seems to work.
Read-only forms do not belong to a <form> element, so we can't refer
to that. Instead find the `.mform` element (which exists for regular
and read-only forms).
While, right now, sites using long (> 10 chars) $CFG->prefix
can continue working (because we still don't have any table
> 28 chars), as soon as some new table with long name is added,
it won't work with PostgreSQL anymore (if the 63 limit is raised).
Hence, this environmental check will verify on both install and
upgrade that the $CFG->prefix is always <= 10 chars.
Sites with longer prefixes will need to rename all their tables
(and maybe other objects, depending on the dbtype) to use a shorter
prefix.
Do not enclose \core_grades_renderer::user_heading() in
$OUTPUT->heading() and enclose the user's name in the user heading in
<h2> tags instead.
- Having <div> inside <h2> results in errors in HTML validation.
- Enclosing the whole user heading in <h2> results in the other
elements in the user heading (Message, Add to contacts) to be announced
to screen readers which can result in confusion.
TinyMCE has a default entity_encoding 'named',
which causes text with diacritic symbols to be converted to HTML entities,
e.g. pâté will be convert to pâté
That will be a problem with the Glossary auto-linking filter if a course has a text concept like pâté,
then the filter will likely fail to get the text concept.
Changing the TinyMCE entity_encoding to 'raw' will resolve the problem.
The token information was not being properly cleared from the $SESSION.
As a result, certain errors were occurring when users attempted to
reconnect within the same session.
When updating a manual enrolment instance, the value of 'notifyall' field must be calculated automatically based on the value of
'expirynotify' field. The relevant plugin's method missed that, thus, the value of 'notifyall' field was never updated when
editing a manual enrolment instance and changing the 'expirynotify' setting.
A functional test has been added to test the entire behaviour of the relevant plugin's method.
When the user picture is being rendered as a link but the user's full
name is not being included in the link and the alt text turns out to be
empty, we could end up with a link without a discernible text. This is
an accessibility issue that will affect screen reader users. If this is
the case, use the full name as the user picture's alt text.
* Add sr-only column headers for the select and actions columns.
* Define unique IDs for the preset name links using array indices.
* Label the radio buttons via aria-labelledby attribute pointing to
the ID of the preset's link.
This prevents automatic casting from int to double on 32 bit systems
(which have a maximum for integers of 2147483647)
Whilst Moodle does not support 32-bit systems any longer, this is still
worth doing to prevent future issues.
In write_close() call update_recent_session_locks() before setting
$sessionatclose so that it includes $SESSION->recentsessionlocks and
doesn't trigger the debugging in check_mutated_closed_session().
Since custom menu items were merged into the primary navigation/more
menu as part of 56c34d71 and related work, the "title" attribute of
each custom menu item was lost.
Ensure that the following both return consistently, so that filepicker
form elements behave the same for each when defining "accepted_types":
'.html, .txt'
['.html', '.txt']
In the description of "Quiz attempt reviewed" events the user IDs of
the reviewer (typically the teacher) and the reviewee (typically the
student) were the wrong way round.
By using a more explicit selector for matching containers we avoid
previous problems such as incorrect count, and collapsing of elements
that had the same class name.
Using the .disabled class for the course module or section that is
being moved does not have sufficient colour contrast against the
background, especially when it is being highlighted.
So instead of using .disabled, just use Bootstrap classes .text-body
and .font-italic which renders the cm/section to be moved in normal
font colour while preserving its italicised look.
* For consistency
* The font colour for .text-muted meets a contrast ratio of 4.69:1
which meets WCAG 2.1 Level AA success criterion for minimum contrast.
This commit also does few things:
- Adds the I change window size step back to avoid future randoms.
- Increase the download limit size on a couple of download steps that
have been failing randomly.
- Adds missing @core_grades to tertiary_navigation_searching.feature.
This commit makes two changes to reduce random failures in behat:
- wrap the toggling of manually configured completion in pendingjs
- set the loading spinner content without forcing the completion toggle
ws to wait for the loading spinner to update.
The problem is that HTML Purifier is not compatible with Markdown,
that means we cannot sanitise Markdown texts before editing.
Luckily Markdown has to use plain text editor which does not have
XSS problems.
The only tiny downside is that Markdown cannot be allowed
in "trust text" areas any more.
Note that, for windows php82, we need to have the php-zip extension
explicitly added. So we are adding it too.
From the manual:
"As of PHP 8.2.0, php_zip.dll DLL must be enabled in php.ini.
Previously, this extension was built-in."
Introduce a lang string `questioniconfollowlangdirection` that will
allow the current behaviour of displaying the question mark icon
in RTL languages in the same manner as the current language.
This fixes the changes made in MDL-75047:
* Change `thisicondirection` langconfig string to a more specific one
like `questionicondirection`.
* Use existing direction string value `ltr` for the value of
`questionicondirection`. Also add a comment to at least provide some
form of documentation for it.
* Improve the logic for applying the horizontal flipping class for the
question icon:
- Check first if the icon is a question icon
- Flip the icon if `questionicondirection` is set to `rtl`.
This was detected only in master (4.3dev) because MDL-40084
added a unit test covering it. But the very same patch is
needed in 402_STABLE, no matter it's not covered there.
Patch makes user dashboard respect permission overrides
that have been set on individual blocks on the system
dashboard (indexsys.php). When a user dashboard is created
either when the user visits their dashboard for the first
time or after an admin reset. When blcoks are copied to
the new dashbaord overriden permissions are also copied.
This is in keeping with other Editors such as Atto and resolves an issue
where DOM Purification incorrectly identifies HTML-like string content
as an HTML tag with a JS variable.
Moodle already performs XSS sanitisation using HTML Purifier in PHP.
Comparison of method void return within a conditional didn't make
sense, and the aforementioned method throws an exception on error
anyway which would halt execution.
This bumps direct dependencies to current ones:
- phpunit: 9.5.x No updates here, there is the 9.6.x series
available but a lot of deprecation warnings have been
introduced there without any alternative to use. See:
- https://github.com/sebastianbergmann/phpunit/issues/5160
- https://github.com/sebastianbergmann/phpunit/issues/5062
- ...
While that will be handy to prepare ourselves to PHPUnit 10
in some months... we cannot force everybody to jump to 9.6.x
because that will make a lot of tests to start emitting
warnings.
So we stay with PHPUnit 9.5.x for the life of this branch.
- mink-phpwebdriver: 1.2.1 No updates here, just changed the
constraint because we cannot advance to 1.3.x yet, there is
a change there causing some app tests to fail. See:
- https://github.com/oleg-andreyev/MinkPhpWebDriver/pull/81
So we stay with 1.2.x until that issue is fixed/clarified,
only then we'll review the status.
- behat: 3.12.x => 3.13.x
And also, automatically, a bunch of 2nd and deepest dependencies.
Generated with php80 that is the lowest php version supported
by this branch and, also, by some of the dependencies, as per
documented @ https://moodledev.io/general/development/tools/composer
Worth mentioning behat/mink-goutte-driver, that we should move
to mink-browserkit-driver, but that's out from this issue scope.
PHP 8.2 has deprecated setting properties on objects dynamically.
The qtype_calculatedmulti question type had two properties being
set this way($correctanswerlength and$correctanswerformat).
This patch extends the question_answer class to add the properties
when the object is instantiated.
The boost theme makes the TinyMCE editor rendered in a scrollable container,
scrolling the editor’s container will cause TinyMCE UI elements to be detached from the anchor.
Therefore, to keep the tinyMCE menu in the correct position,
adjustments must be made on the page drawers style.
When using TinyMCE editor in Safari browsers, a problem may occur where the dialogue
windows (such as modal dialogs) overlap with page drawers due to a specific behavior
in Safari's rendering. This function addresses the issue by adjusting the CSS overflow
property of the page drawers, ensuring they do not obscure the dialogue windows.
* Using the generator to create the forum discussions, the timecreated
of a discussion can be set before the user's last course access which
will prevent forum_print_recent_activity() from including this generated
discussion in the recent activity results. To work around this, generate
the forum discussion 1 second after the current time to make sure that
the user's last course access will always be before the discussion's
creation time.
* This patch also includes some optimisations by:
- Removing the unnecessary @javascript tag for the
`Time limit expires` scenario
- Bringing the discussion generation to each scenario to allow each
scenario to customise the data passed to the generator like for the
`Time limit expires` scenario.
- Navigating directly to the forum instance.
This commit does few things:
* Replace manual forum posts and replies to use data generator.
* It also changes the code to handle social forums.
* Other behat clean-ups and optimization.
Co-authored-by: Simey Lameze <simey@moodle.com>
This commit also does other things such as:
- Replace manual steps to setup gradebook by data generators
- Added support for outcomes and scales page resolvers
- Behat clean up
The current I am on course homepage with editing mode loads twice the
course page. This patch prevents this from happening by going directly
to editmode.php.
Before this change conversion tasks would be queued for assignments that
did not have the pdf annotation method enabled.
The change will also prevent queing when the plugin is disabled.
Aurora MySQL does not support row format COMPRESSED and falls back
to COMPACT if you try to use it, cuasing column size too large
errors if you try to use it with utf8mb4 collation.
When libtidy removed option 'drop-font-tags' the docs said to use the
'clean' option instead (see MDL-69951). However, 'clean' replaces
legacy style information with classes, the corresponding CSS added in
<head>. With 'show-body-only' (used because we're inserting an HTML
fragment mid-way through <body>) that <head> with its CSS is omitted.
So 'clean' with 'show-body-only' removes legacy style information,
rather than tidying it. Consequently we now drop the 'clean' option.
This means <font> tags are no longer dropped as libtidy currently has
no equivalent to the removed 'drop-font-tags' functionality.
Declare qformat_xml::answerfiles and qformat_xml::feedbackfiles instead
of dynamically declaring answerfiles and feedbackfiles for the
question_answer instance when exporting questions and answers to XML
format.
The question answers for question type plugins extending
\qtype_calculated now have the type
\qtype_calculated\qtype_calculated_answer. So the answers for test
helper functions make_calculatedmulti_question_singleresponse() and
make_calculatedmulti_question_multiresponse() should also be changed
accordingly.
The question answers for question type plugins extending
\qtype_calculated now have the type
\qtype_calculated\qtype_calculated_answer. So the answers for
make_calculatedsimple_question_sum() should be changed accordingly.
* Question type plugins that extend qtype_calculated should be made
aware of the changes about the new question_answer type
\qtype_calculated\qtype_calculated_answer.
Where only the case of the idnumber field was being updated, we
triggered an exception because checking for an exising idnumber
was not excluding the current category (i.e. the one being updated).
This commit addresses compatibility issues in the qtype_multianswer
question type. The following changes were made:
- The usage of the undeclared property $maxmark on subquestions in
qtype_multianswer was identified as problematic. To resolve this,
the existing declared property $defaultmark is now being used for
the same purpose.
By switching to the declared property $defaultmark, compatibility with
PHP 8.2 is ensured and the issue with the undeclared property is
resolved.
This commit addresses compatibility issues in the qtype_calculated
question type. The following problems were identified and resolved:
- qtype_calculated was reusing the qtype_numerical_answer class and
setting two additional properties ($correctanswerlength and
$correctanswerformat) that were not declared. To resolve this, a new
class, qtype_calculated_answer, was defined and utilised.
- During grading in qtype_numerical, there were references to an
undeclared property ($unitisright) used to track calculation details.
This issue was fixed using a 'replace temp with query' refactoring.
These changes ensure the qtype_calculated question type is compatible
with PHP 8.2 and addresses the identified issues.
This commit addresses the following issues and improvements in the
qtype_truefalse class:
- Added missing class properties:
- public $truefeedbackformat
- public $falsefeedbackformat
- public $showstandardinstruction
- Updated qtype_truefalse_question to include type hinting for better
IDE support.
Otherwise the enrolment index page tries to load the course index
drawer, which doesn't exist at this point and triggers exceptions in
the editor module.
Co-authored-by: Tim Volckmann <tim.volckmann@mastersolution.com>
The completion_rule_enabled() method was not checking all the fields
and, in some cases, it was returning false (because it was not checking
that "Count of entries" field was enabled).
Change the param type of the category name when editing to allow
for entry of multilang tags, the formatting of which is already
supported when returning category name.
Co-authored-by: Hiroto Kagotani <hiroto.kagotani@gmail.com>
Ensures cross-DB compatibility for all types of both custom and user
profile fields, specifically when they're filtered and/or aggregated.
Implement stress testing of both via appropriate report sources.
The selected value of the dropdown was being cleared when the dropdown
was opened, this was causing an error when submitting the form if the
field was required. The main cause was that we were emptying the
select before loading new values, this solution was created because
of an issue where we could not deselect values if the list was
reloaded. To fix this problem, I added an empty option as the first
element of the select only when deselecting a certain item.
If we cannot resize the given file (files such as ICO can't be, for
SVG it makes no sense), then we should just return the original file
as is rather than an error.
From my experimentation I have discovered that Safari does not properly
respect the standard MediaRecorder mimetype. Instead of using `codecs=`
it uses `codecs:`.
This change copies the codec array to have the list of possible codecs
include both codecs= and codecs: whilst remaining in order.
Add a new method for activity_header that determines the heading level
depending on whether the theme displays a heading for the activity
header (usually a h2 heading with the activity name).
E.g. in Boost, the activity name is already being displayed in a
heading. So page headings can be rendered as h2. However, on Classic,
the activity name is being displayed as a h2 heading. So headings need
to be adjusted for the activity pages.
- Improve the grammatical structure of the lang string.
- Remove the mention of "link", for sighted users, it should already
be obvious that it's a link. For screen reader users, screen readers
already announce links. So it would be confusing for them to hear
something like "Click 'Moodle.org, link' link to open resource"
Add a help text explaining that the URL resource's name field will be
used for the link text. Also add basic guidelines about naming links for
the benefit of screen reader users.
LTI 1.1 uses OAuth 1.0, which requires 'hostname', among other things,
as part of signature verification. When Moodle is run behind a reverse
proxy, $_SERVER['SERVER_NAME'] is an unreliable way to fetch hostname.
It isn't always equal to the external hostname but instead can contain
the local host, meaning the signature verification will fail. This
change replaces the use of $_SERVER vars with $FULLME - which already
takes this into account - to fix the signature verification.
Co-authored-by: Berengar W. Lehr <Berengar.Lehr@uni-jena.de>
It turns out that there are plugins which don't (currently)
clean up their question attempts when a context is deleted.
Therefore, we need to make Moodle core robust to that.
This covers the case where a course is published and the launch data
doesn't include the 'lineitem' property of the ags claim, meaning the
tool can manage its own line items.
With courselistshortnames enabled Moodle 3.11 and earlier showed
"Category | Short name" on the first line of course cards (for example
in block_myoverview). From Moodle 4.0 the category is moved to the last
line making the first line "Short name |". This change removes that
stray "|".
* Process markdown and other allowed formats (see FORMAT_XXX) in get_label_name
so it does not display it as a litteral string in the course index menu.
This handles things like site policies, which store the current URL,
redirect to the policy agreement, then redirect back the current URL
afterwards. In such cases, we want to redirect back with 'launchid' set
so that we can fetch the id_token from the session cache. This is the
same thing we already do during account binding, so the patch only
makes sure the PAGE->url is properly set before calling require_login.
Now that it has been decided to backport php82 to Moodle 4.2.x, we
can already fix the php versions to be checked by the coding
standard (they were previously open for the branch).
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In these cases, mod_assign was treating other unrelated objects as its
personal object store. These have been updated to not require this.
Ensure we don't create nested <script> tags when rendering templates,
which triggers console errors since 1ef815cd.
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
Changes done to fix the warnings thrown in the history view when the
last version of a question is deleted and improved UI to display the
message and a Continue button, to navigate back to the Question bank.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
At the test_transaction_with_events() I changed $this->_called into $called local var,
since the $called is used inside the anonymous function I added the third param as
a variable passed by reference so the last assert can get the newest value.
Replace steps that manually add page instances via the UI and use
Behat generators. This improves the speed of Behat test runs.
Co-authored-by: Simey Lameze <simey@moodle.com>
The property was added to the config options in c3f4123b, however no
default value was provided, meaning if it was ever unset then a PHP
type error would be triggered on the null return.
If an activity, like workshop or forum, has multiple grade items,
declarative binding of the grade item (line item) isn't supported.
Instead of throwing an exception, handle the case more elegantly
and just omit the 'add to gradebook' option for these activities.
Converting these tests to use an onlinetext submission and drop the JS
requirement means that they are much less susceptible to runtime issues,
and not susceptible to those caused by JS at all.
This random failure was happening because of the time required to render
and interact with the file picker which sometimes exceeded 2 seconds,
depending on the speed of the machine running the tests.
Converting these tests to use an onlinetext submission and drop the JS
requirement means that they are much less susceptible to runtime issues,
and not susceptible to those caused by JS at all.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
If a statically accelerated cache returns an empty array then the value
was still fetched from the non-static cache store.
The check of the `$result` should be strictly checked against `false`,
which is the value used if no value was found.
When creating a course, a manual enrolment instance is added by default. The instance settings should inherit the values of those
for the manual enrolment plugin and properly calculate its extra settings. The 'expirynotify' setting wasn't inherited correctly
in case it had 'Enroller + Enrolled' value.
A functional test was added to test the behaviour of settings inheritance.
For backpack connection using OBv2.0, email and password can't be
blank.
This patch adds some extra checks to validate this and displays an
error to the users before trying to connect to the backpack.
Besides, the error displayed when any error is returned by the
backpack (like invalid credentials), has also been improved, to
make it clearer for the users.
The missing version bump caused cache configuration to not be updated,
and the static acceleration cache size was therefore not updated. The
result here was that the static acceleration cache was constantly
emptied leading to an increased page load time.
Import now checks whether assignment is an group submission and
matches group ids instead of user ids in that case.
Additionally check for original unchanged files to not import in
group files instead of user files.
MDL-45580 introduced the readonlypages filearea, and when loading
page images for an attempt, the code would check if the pages existed
, creating them if not. The code inside this block also contained
a guard clause for the case where no readonly pages existed - which
is a situation that should not happen. Whenever readonly pages are
requested, they should exist.
MDL-66626 introduced a situation where page counts not matching would
also retrigger page generation. However this led to a situation where
the guard clause could be entered when requesting readonly pages.
This patch refactors the guard clause, and improves the logic to
regenerate pages.
When requesting the readonly version of pages (which contain
teacher annotations), they should always be available - the
PHP side even throws an exception when they're not. This
means we don't need to worry about polling document converters
from the JS side and can just return the pages immediately.
The behat step "I draw on the pdf" creates the exact same line every
time it is called. This can cause issues in a situation like the following:
- Student uploads PDF
- Teacher annotates page 1
- Student edits submission, adding another PDF
- Teacher annotates page 1 again
- When the student looks at their annotated PDF it appears the same as
before they added the second PDF (i.e., they don't see extra pages for
the second PDF)
Exactly why this happens isn't clear, and in practice this probably never
happens because the chances of drawing the exact same line twice are almost
0. However, when testing the functionality added in MDL-45580 it was causing
issues.
Only call user_update_user when the relevant user data has changed,
preventing unnecessary user_updated events. This also removes the
line setting timemodified on the user since user_update_user already
handles this.
When receiving data from the LTI launch, or service call, only update
the user record when we know something has changed. This prevents the
creation of many \core\event\user_updated events.
This isn't required and generates excess events. The only thing update
does, other than update the user fields, is to update the picture, but
in this case, that's unset in membership-based auths anyway, so it's
entirely safe to remove this.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
This is as regression of MDL-72413, when description was moved to a
standard module API.
In some cases, when description has only images, it's was not
displayed.
This is a regression when the activity pages were redessigned in Moodle
4.0. The default introduction should be displayed when the activity
description is empty.
This commit does few things:
* Unify data generators usage to role short name.
* Replace remaining manual steps to use the new data generator.
* Also replaced other manual steps to set config to use data generators.
* Tidy up of some tests, aligning pipes and splitting one line steps into multiple lines.
* Fixes tests to have one Given/When/Then per scenario.
If a plugin attempts to do something that would lock coursemodinfo cache
during install (such as creating a course module), this currently fails
as check_lock_state is not implemented in cache_disabled. Adding the
cache_loader_with_locking interface ensures that all lock methods are
implemented.
If the content in Atto is wrapped by an element with only a background color, without text color,
the accessibility checker will use the parent text-color as the foreground color, which is owned by form-control,
it becomes a problem if the wrapped element background using #34274f and the parent text-color is #495057 then
the low contrast as a result.
To avoid such a situation, the patch will only calculate color contrast if only the element has text,
so it will give results per element if the accessibility checker is unsatisfied.
If the content in Tiny is wrapped by an element with only a background color, without text color,
the accessibility checker will use the parent text-color as the foreground color, which is owned by form-control,
it becomes a problem if the wrapped element background using #34274f and the parent text-color is #495057 then
the low contrast as a result.
To avoid such a situation, the patch will only calculate color contrast if only the element has text,
so it will give results per element if the accessibility checker is unsatisfied.
The patch also changes the selected element from 'body > *' to 'body *',
so it will return all elements (parents and children) in the content instead of parents only.
And minor modifications on the warning_content.mustache to fix the numbering format.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
To make Moodle 4.2 compatible with PHP 8.2 regarding the deprecation of dynamic properties,
several changes have been made from version 5.22.5 (https://github.com/adodb/adodb/compare/v5.22.4...v5.22.5) and
then applied the changes to Moodle 4.2.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Removing the existing $id property at portfolio_format_leap2a_writer, because
the $id only needed inside the __construct().
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Adding @runTestsInSeparateProcesses and removing setUp() and tearDown() at the
component_test.php, so the test will run in a separate process making sure that
whatever changes happen to the in-memory version of the component class in its thread,
therefore does not impact other tests and makes it more safe.
This commit makes the following improvements to mod_imscp tests:
* Removes unnecessary @javascript and @_file_upload tags from non-JS tests.
* Removes user/enrol data generation from tests that can be performed as admin.
* Removes "I log out" and other unnecessary steps.
In this commit, the following improvements were made to the mod_lti Behat tests:
* Replaced manual steps with data generators to set completion.
* Eliminated unnecessary user and course enrolments data generation as some tests can be performed as an admin.
* Removed the @javascript tag from non-JS tests.
* Updated the LTI data generator to generate an internal Moodle URL in the toolurl field, enabling the use of XML files.
This feature had lots of small issues and it made sense to fix it whilst
investigating a query:
* most of the steps do not require JavaScript
* it uses the UI to set an admin setting, for every scenario:
** only 3-4 of the scenarios actually test that setting
** it is very slow to do it his way when we have a generator step we can use
* we create two assignments in the Background, but each test only uses one of them
* we create the assignments in the Background with a generator, but
update them to modify various settings in each Scenario using the UI
when we should just create one assignment for each test and set it up
correctly for that Scenario
The feature added in 4.1 to lock the modinfo cache does not work when
using Redis, because:
* The API to acquire a cache lock is confusing, and the code did not
check that it successfully acquired a lock before going on to build
the cache anyway.
* Unlike the other types of cache lock, the Redis store did not retry
the lock for a timeout period before giving up and failing.
This change fixes both points.
It is perfectly legitimate to create and/or use a method named `define`
in JS outside of RequireJS.
Unfortunately our requirejs.php wrapper is dumb and does not understand
this.
In the long term we need to stop doing this at all. We really should be
able to already, but every time I try to something prevents it.
In the interim, this change adds a secondary check to see if there is an
existing define which _does_ have the right name in it already.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Restoring multiple quizzes from a pre-4.0 backup was broken when
the quizzes shared a random question. This is because after the
first quiz created a set reference in place of the random question,
it deleted the question record so it was not there for the second
quiz to use. This change tracks the IDs of random questions so
they can be deleted at the end.
Several version checks were incorrectly using
restore_controller::info::moodle_release instead of moodle_version as a
version number. This replaces all of those checks with a common pair of
methods to make the checks clearer and more maintainable.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
// Removed test_add_malformed_favourite()
PHPUnit version >= 9.5 no longer converts PHP deprecations to exceptions by default.
To comply with PHP 8.2, which will deprecate the dynamic properties,
we need to remove test_add_malformed_favourite()
because the test will raise a warning for dynamic property deprecations.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Course date registration data reports correctly.
MDL-77306 added a count of how many courses there are
without start and end dates. However, all courses have
a start date, it is only the end date that is optional.
This patch fixes the data collection and makes the lang
string describing it clearer.
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).
Previously, a set of calculated quiz statistics would only 'last' for
15 minutes. Then they would be considered invalid and not used.
Now, computed statistics are kept indefinitely. Instead, when a new
batch of values are computed for a particular set of settings, older numbers
for the same settings are deleted first. Therefore,
question_stats_cleanup_task is no more.
Previously, when users viewed the question bank, if the quiz statistics
had not already been calculated, we would try to compute them there an then.
This could be very, very slow, leading to session lock problems.
Now, we never try to compute the statistics on the fly. Instead, we rely
on the existing \quiz_statistics\task\recalculate scheduled task to do it.
There is a new method on the column class, for the column to declare any
statistics it requires. The question then efficiently loads all those
statistics in one go, and makes them available as the columns are
rendererd, or do any subsequent processing.
According to Boostrap Popovers documentation for popovers triggered
on focus:
"For proper cross-browser and cross-platform behavior, you must use the
<a> tag, not the <button> tag, and you also must include a tabindex
attribute."
This test apparently fails sporadically on Oracle, without obvious
causes. The failure however was hard to interpret, and the test
itself made a lot of assumptions about the system re. naming of
categories/courses. This could be prone to errors, so improve that.
When the H5P third-party library was upgraded to moodle-1.23, the calls to
libraryToString(xxx, true) where replaced with libraryToFolderName().
We missed this, which was doing the H5P Timeline activities weren't
displayed properly.
Thanks Joseph for raising this and preparing a patch! Well done!
Co-authored: Joseph Rézeau
Recent additions to timezones from the IANA TZDB are not necessarily
reflected in those used by the intl extension and/or ICU zones:
* America/Ciudad_Juarez
* America/Nuuk
* Europe/Kyiv
* Pacific/Kanton
Create mapping to allow these zones to be used without errors.
The apiBase in .well-known/badgeconnect.json was ignored and it was
causing some failures when connecting or sending badges to an
external backpack.
For OBv2.1, it has been changed to always use the apiBase defined
in the badgeconnect.json backpack provider.
Right now the order of the groups returned by groups_get_user_groups()
is not fixed and, every DB, can return them in any order. While 99%
of times the order will be the creation one, depending of the RDBMS
decisions, they can be returned in any order.
This is specially noticeable with Oracle, but can affect to any DB.
And that makes some tests (expecting a given order) to fail.
This commit fixes the problem by making the order of the groups
deterministic, because that's the way we use to fix these problems.
Alternative is to relax the tests so only values are asserted by
using assertEqualsCanonicalizing().
If the enrolment instance (the 'published resource') has been upgraded
from LTI 1.1/2.0 to LTI 1.3 (i.e. a new instance was not created),
prevent legacy launches which may occur from old resource links. Only
LTI Advantage launches should be permitted through the method.
If the enrolment method is updated from an LTI 1.1/2.0 tool to an LTI
1.3 tool, it may have associated enrol_lti_users records not having
ltideploymentid values. These are legacy users and must not be returned
by the repository, which deals only with LTI 1.3 LTI users.
Random failures after manually adding block instances (specifically
on slow/Oracle systems) can be entirely avoided by using the block
test generator for scenarios that are not specifically concerned with
manually adding blocks.
While the uses here were not causing too much problem with PHPUnit
execution in any OS, better have them explicitly closed to better
serve as reference for other, future uses anywhere in codebase.
Always that a stream is used, better we close it as soon as possible,
not relying on destructors, gc and friends.
And do it as soon as it's not needed any more. Else, only
_destruct() / shutdown / gc will, and that can be problematic
if many streams are open, or, under some environments like
Windows that consider the stream underlying files still busy,
because they have not been closed.
Although the screenshots in the failures for some of the scenarios in
filter/displayh5p/tests/behat/h5p_filter.feature, like "Render a local
H5P file as teacher" were displaying the expected result, there were
some javascript errors (probably due to behat is quicker and the
iframes were not always ready).
I've added one extra step before accessing the iframe to give more
time to the H5P player to load and confirm the page is displayed
properly.
* The object returned by update_question is alwasy a new clone
and the $question passed in will not be modified.
* The returned object has the fields like questionbankentryid and
the ones related to versionning, so it is more like the data
returned by question_bank::load_question_data.
The used the exist in Moodle up to 3.11, but then was removed with
insufficient thought in 4.0 (because we had grander long-term plans
which still have not happened). Until those plans happen, this
commit adds the simple link back on the preview screen.
When the two restore forms for searching courses and categories were
converted to core templates in eb9935c9 they lost the named submit
button, which broke searching.
The widgetbase module was intended to be a generic search widget,
despite its location in grade/amd/src. As such, other modules may
depend on this. This was modified in MDL-76246, which added new
requires params to the js, and changed the js to expect certain new
data attributes in the templates. This broke b/c for existing
dependents.
This patch makes sure the existing uses of the basewidget continue
to work by adding b/c code. See MDL-77991 which deals with deprecating
this search widget and replacing it with one that just implements the
desired combobox logic, without the b/c code. That issue should also
make it abundantly clear that this widget is for public consumption.
When we deprecate the use of a file, we often include tests which ensure
that the legacy behaviour is maintained. There are also legacy uses
in the community where people would like to use the deprecated API for a
period.
The issue that we face is that, if the deprecated file is included once,
then it will be included for all other, unrelated, tests. This means
that other tests may not detect cases where the deprecated file was
included.
We can solve these cases by running the test that performs the inclusion
in a deprecated process. This means that the inclusion is only performed
in that isolated process, and other unrelated tests do not include the
file.
However, we also then need to detect which files which are including the
file and which we do not know about.
This change introduces:
- an override to the TestCase::setInIsolation method to define a
constant when the test is running in isolation
- a new function that a file can call when it is included, to make sure
that the test process was isolated, where there is any test.
Move the fixed assignment removal upgrade code to the end of the upgrade
script as a new upgrade step. Already upgraded sites with mod_assignment
removed should just be able to rerun this without any issues.
When an editor is renderer initially invisible to the browser, e.g.
the forum "Add discussion" form, it has a `clientHeight` value of
zero. We can approximate an alternative value based on the number
of rows in the textarea.
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
Previously the content of the stored file was extracted and stored in a
variable, passed around, and then submitted to MoodleNet. This results
in very high memory consumption when dealing with MoodleNet.
The stored_file should be passed around as a first-level param to
discourage this, and the content should _never_ be loaded into memory.
Instead file streams and resources should be used to allow Guzzle/Curl
to buffer the file from disk/other storage straight to MoodleNet.
This change introduces a new get_psr_stream() method to:
* stored_file
* file_system
This allows us to fetch a Psr Stream implementing the
PSR\Http\Message\StreamInterface and pass it into Guzzle, which means
that there is no need to load the entire file content into memory to
serve it.
We shouldn't invent fileareas with invalid values (specifically in
this case, the `itemid` overflowed maximum length of the field on
Oracle).
Given the generated file is transitory in nature, instead store it
in the users draft area.
It's possible to have some Moodle components soft linked instead
of being real directories within codebase (within dirroot).
For example, Composer's "vendor" directory can be soft linked
(from elsewhere), or also plugins can be installed using soft
links.
In those cases, Moodle calculates the $SCRIPT global as null. And,
then, string operations on it are emitting a PHP deprecation message
with PHP 8.1 and up.
This fix just ensures that the behaviour is the same than before
PHP 8.1, aka: ltrim(null) = '' (empty string), without any PHP warning.
This makes it possible to share a single activity to a MoodleNet
instance (which has been enabled for sharing in site admin).
This utilises the core\moodlenet\activity_packager to create an
activity backup, then sends it using the MoodleNet create resource
API specification.
Originally implemented as MDL-75932
This makes it possible to package a single activity as a Moodle backup
file, excluding user data so it is ready to be shared as a MoodleNet
resource.
Co-authored-by: Michael Hawkins <michaelh@moodle.com>
Originally implemented as MDL-75932
The icon's iconclass context data adds additional CSS class(es) to
calendar event icons to better control how the event icon is displayed.
e.g. without filtering for activity events that don't hae monologo
versions of their icons.
Ensure the same heading tags are available as those defined in the
editor block formats configuration (c51b7e2c).
Co-authored-by: Andrew Nicols <andrew@nicols.co.uk>
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.
In MDL-77086 some extra information was added when a badge can't be
sent to a backpack. However, as the Badgr is not following the
specification, it's still hard to debug it when it fails (as
raised in MDL-75552, Badgr is not working because, in some cases,
it's returning the "Method Not Allowed / Request method 'POST'
not supported" error.
In some cases, $CFG->lang might not be set, and this is causing a
Notice to be displayed when, for instance, database connection fails.
This patch should fix this case.
- Only resize if the H5P EmbedCommunicator is defined (otherwise, it was causing a
JS error)
- An unnecessary image has been removed from the greeting-card.h5p fixture package.
That way, the text will always be displayed (even if the iframe is still not
resized). Instead of replacing the original greeting-card-887.h5p file, I've
renamed it to greeting-card.h5p, to remove these ugly and unnecessary numbers
at the end of the file name).
To temporarily prevent the PHP 8.2 warning about Dynamic Properties' deprecation,
the #[AllowDynamicProperties] property was added. We encourage anyone to perform the upgrade
as soon as the Google library complies with PHP 8.2.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
While working on this issue, even though it's not related to the deprecation issue,
I added 'use core_external\external_api' to avoid the 'Undefined type external_api' warning at
mod_lti_testcase::generate_tool_proxy().
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
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).
When questions are filtered by tags in the question bank, the qtagids
params are passed in the array format. Though moodle_url handles this,
single_button::export_for_template cannot. Hence changes done in
weblib.php to provide params for export_for_template in the
suitable format.
Thanks Huong. I have added the Behat test you provided in the patch.
The YUI Overlay widget encloses the subquestion feedback in a div
which causes a div element to be enclosed in the subquestion span. This
leads to an accessibility issue in terms of HTML parsing as inline
elements (span) should not contain block elements (div)
The YUI Overlay widget is also not accessible as it does not really hide
the overlay contents via aria-hidden when the overlay is not being
shown. It's better if we stop using this and use Bootstrap's
popover component which is more accessible by default.
This patch also removes module.js for the qtype_multianswer plugin as
it only contains codes related to rendering the feedback contents in the
YUI overlay widget which is no longer necessary.
- Table cells are top-aligned
- Cell paddings changed to 10px
- In the header row, the user detail cells don't wrap but other cells do
- Grade item names wrap when they are long
- Action menu button for the grade cells does not go to the next line
when the cell is not in the header row
- Heading cells in the footer row are right aligned
Use native ANSI SQL syntax for numeric comparisons where possible,
define filter API for the case where filters must re-use the given
field SQL while ensuring uniqueness of any field parameters.
Currently only necessary in the category filter type.
* CLI adhoc_task.php: new option --taskslimit
* CLI adhoc_task.php: --execute option is implied when other options are given
* core\task\manager::get_next_adhoc_task(): new param $number
for limiting number of tasks to run
* CLI adhoc_task.php: new option --id
* cron::run_adhoc_task($taskid) for running tasks by id
* core\task\manager::get_adhoc_task($taskid) for retreival/locking
The restored states should be treated differently, to avoid resetting
them by default the first time users access them.
In that case, the first time users access to content with restored
xAPI states, they should be displayed (instead of resetting them).
The 'Content state is removed when an attempt is created' scenario
is failing randomly because the content for the student2 was not
always saved. Some extra steps have been added to give more time
to the player to save contents and also to check it has the
expected value.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
This adds "core_question > updated question" as an entity for `the
following "X" exist` and calls the existing update_question() generator
which will create a new question version with the supplied data.
This works by always outputting both the customised and uncustomised
number into the HTML of the Questions page, then uses CSS & JS to
show the appropriate one as shuffle is toggled.
The `filename` field was only used by each to determine whether the
file was itself a directory, and it's presence meant that aggregation
of each column wasn't working properly.
Adding LIMIT to the primary auth type SQL statement does not play well
with Oracle. We can just remove the LIMIT clause from the SQL statement
and and pass IGNORE_MULTIPLE to the call to $DB->get_field_sql()
instead.
When a new attempt is created, the previous state content is removed.
There was a bug which was removing all the content data of a given
activity for all the users (instead of deleting only the content data
for the current user).
To prevent future regressions some extra checks has been added to the
bulk_select feature file. The new checks validate the activities and
sections bulk selectors are really selected, not only the total count.
AMOS BEGIN
CPY [type_contentbank,core_plugin],[type_contenttype,core_plugin]
CPY [type_contentbank_plural,core_plugin],[type_contenttype_plural,core_plugin]
AMOS END
When fetching the list of tasks to run, we were using the _process_
start time as a base. The various task fetchers (adhoc and scheduled)
only fetch tasks which were due to run _before_ this time.
With the new keepalive setting, we need to pass both the process start
time, and the run start time. We need to use the process start time to
determine whether MUC caches have been cleared since the process
started. but the run start time to fetch recent runs.
The Assignment 2.2 activity module was disabled in 2012 but kept to
enable pre-2.2 backups to be restored and have the assignments
auto-converted to new assignments. After almost 10 years, it's time
to remove it from Moodle core.
After checking the the relation between mod_assign and mod_assginment,
it was found that upgradelib.php was only used/called from the
mod_assignment. Therefore, it was decided to move the upgradelib.php
from mod_assign to mod_assignment and deleted the line in which requires the upgradelib.php
* Recording formats should not be accessible to students unless specified as safe
* A global setting will now list all the "safe" formats for students
* A new capability 'mod/bigbluebuttonbn:viewallrecordingsformats' will allow users to
see all formats (as long as they can manage recordings also)
This test is adding table rows into an HTML editor and, quite rightly,
is failing under TinyMCE.
This only worked under Atto because Atto does not perform any HTML
validation. A user would not actually be able to insert the template
content this way, and therefore the test is/was invalid.
This change disables the HTML editor before inserting the template
content, which resembles the actions of a real user.
This install script has not been required for a number of years because
we require the previous LTS for an LTS+1 upgrade. Once Atto was
part of core after an LTS release, it was safe to remove this file.
This is now preventing Tiny from becoming default for new installs in
some situations.
* The file file_remove_editor_orphaned_files should take into account URL that
have been embedded in a tag content instead of an attribute (like src attribute)
* This will fix issue with inserting H5P content in calendar events.
Add the DB type, count of courses with start or end dates set,
default site theme, and primary auth type statistics to those
collected as part of the site registration data.
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.
This is a workaround for an upstream bug which I have not been able to
reproduce outside of Moodle whereby the editor.contentWindow does not
match the editor.iframeElement.contentWindow when it should.
This issue only seems to affect Firefox, and it may even be a bug in
Firefox. It can only be reproduced when using a fresh browser which has
never had a TinyMCE window open.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
When rendering content items, check whether the plugin has monologo
icons. If so, add a 'nofilter' class so the plugin icon can be
rendered as is and without the CSS filter.
Add the `.nofilter` class for activity icons when the icon URL's
`filtericon` parameter is not set, so they get rendered as they are on
the timeline block.
Add the `.nofilter` class for activity icons when the icon URL's
`filtericon` parameter is not set, so they get rendered as they are on
the recently accessed items block.
Add the `.nofilter` class for activity icons when the icon URL's
`filtericon` parameter is not set, so they get rendered as they are on
the context header on the activity page.
Add the `.nofilter` class for activity icons when the icon URL's
`filtericon` parameter is not set, so they get rendered as they are on
the course homepage.
* Apply the filter CSS property only to activity icons
that don't have the ".nofilter" class. This will allow
activities with non-SVG icons to be rendered as they are.
* If a plugin defines a `filtericon` custom data or uses its monologo
version of the icon, a `filtericon` parameter is being added to the
icon's URL. This information can help plugins determine whether to
render the activity icon as is or with CSS filtering.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
At certain cases I didn't see a tick in view mode. For example
when I click grades only and then switch to totals only a tick
next to total only was not shown. I think this is caused since before
mode was changed in circlurar order i.e.
fullmode->grade only->aggregate only->fullmode. But now any order is
possible so we need to respect it
The step i_enable_plugin cannot be used as bigbluebuttonbn_default_dpa_accepted
setting needs to be enable in order for the BigBlueButton plugin to be enabled.
* Unify the fields_action_bar and action_bar template
* Make sure the Create a field button is a primary button
* The fields_action_bar is very similar to the action_bar template so
should be used instead
Not critical to have, but better don't mention it. Note that all
the cases need to stay because there may be other environments
where the same problems are reproducible.
Only exception is the CURL_SSLVERSION_TLSv1_2 constant, that
can be safely removed because it's available since PHP 5.5.19.
Agreed @ MDLSITE-7135, this completely removes the Travis CI
integration with a note in the lib/upgrade.txt file, pointing
to the currently supported GHA alternative.
* get_join url should not have as many parameters as they can be deduced from the instance.
* Refactoring it will allow to pass on more implicit parameters like the user profile file.
MDL-69087 added an option to let themes personalize the H5P styles.
This patch implements the h5p_alter_styles() method in boost, to
let admins personalize the H5P styles using the 'Raw initial SCSS'
and 'Raw SCSS' theme settings.
That way, users won't need to create their own themes to define
some styles for the H5P player.
- It was a mistake to assume the listbox is always within
combobox.parentElement
- Take into account that the popup of the combobox is not necessarily a
listbox
- Update combobox fix so that it also work with comboboxes that are not
select-menu
- Update combobox fix so that it also support editable comboboxes
- Update click listener to take into account that the event's target
might be one of the option element
- Having a hidden input element for comboboxes was not an ARIA
requirement and was added by us. I added data-input-element to the
combobox element to specify the input element related to it.
Ensure that when the user entity is added multiple times to a report,
when there are custom profile fields, each of those gets a unique table
alias per-entity.
Evaluated usage of "Install selected language pack(s)" in Behat and
replaced the steps to use generator to install language packs as part
of test setup.
- Filter searchable courses if set.
- Sites with large groups should now experience better performance.
- Applies and optimises query against a proper date range.
Co-authored-by: Mark Sharp <mark.sharp@solent.ac.uk>
Co-authored-by: Peter Sistrom <petersistrom@catalyst-au.net>
Co-authored-by: Kevin Pham <keevan.pham@gmail.com>
Most section related actions require the frontend to use alternative
strings depending on the format plugin lang file. This patch adds
overridden strings to the current setViewFormat course editor setup
object.
The new Font Awesome version needs some changes in the way icons
are displayed from SCSS files:
- The atribute 'content: "\xxx"' needs to be converted to
'content: fa-content("\xxx")'
- The style must be declared. The regular is used by default, so all the
solid icons needs to add '@extend .fa-solid'
- @include fa-icons() is not required anymore (if used, the icons
are not displayed properly).
While working on this issue MDL-76931(Upgrade core-js-bundle),
we need to convert a HTMLCollection like .children to an array to use forEach(),
so we can upgrade the core-js to the upper version.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
After upgrading the Font Awesome from 4.7 to 6.3.0, the moodle.css
file is bigger than before, so a few behat tests related to the data
export (where the moodle.css file is included), need to be updated
to meet the new size.
This patch updates the Font Awesome library version from 4.7 to 6.3.0.
The readme_moodle.txt has been updated properly because the new version
works slightly different.
There is a file to add backward compatibily to version 6 files (shim).
More information about how to upgrade from 4 to 6 can be found in
https://fontawesome.com/docs/web/setup/upgrade/upgrade-from-v4
Although a few icons have been renamed in Moodle 6, as the new version
is compatible with v4 icons, for now the references to the old icons
(fa fa-xxxx --> fa-solid fa-xxxx) haven't been changed yet.
A separate issue will be created to update them in the future and
review if some new icons can be used too.
Following MDL-65835 this now finally deprecates and removes the
functions:
- plagiarism_save_form_elements(), please use
{plugin name}_coursemodule_edit_post_actions() instead
- plagiarism_get_form_elements_module(), please use
{plugin name}_coursemodule_standard_elements() instead.
- plagiarism_plugin::get_form_elements_module(), please
use {plugin name}_coursemodule_edit_post_actions() instead.
- plagiarism_plugin::save_form_elements(), please use
{plugin name}_coursemodule_standard_elements() instead.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
In the quiz reports, for any 'Never submitted' attempt, there is
now a 'Re-open' button next to where it says the attempt state.
If that is clicked, there is an 'Are you sure?' pop-up. If the user
continues, the attempt is reopened. If the student now has time left,
the attempt is put into the In progress state. If there is not time
left the attempt is immediately submitted and graded. The
'Are you sure? pop-up says which of those two things will happen.
This allows an availability plugin to indicate that a selected condition
should be considered "private", and therefore should never be shown to
users who don't satisfy the criteria. The availability_groups plugin
uses this to protect visibility of groups that are not visible to
non-members.
These new settings are designed to enchance user privacy surrounding
groups. They allow groups to be configured so that users outside the
group cannot see the group, so that users in the group cannot see each
other, or so that users cannot see the group at all, even if they are in
it. This avoids issues where a group may be assigned based on sensitive
personal information (such as a person requiring special arrangements
due to a disability).
By default, groups are visible to all and available for participation in
activities, which maintains the current behaviour.
For performance, a new cache has been added to track the number of
groups on a course that are not visible to non-members. This allows us
to revert to the existing behaviour if the new features are not being
used at all on a course, and only apply the new visibility conditions if
they are.
Users who have the moodle/course:viewhiddengroups capability should be
concious of exposing hidden groups when showing their screen to other
users. The "Switch role to..." feature can be used to show a course page
on screen without exposing private availability conditions, for example.
The changes cover several specific areas:
* grouplib functions, which most code should use to get lists of groups
and members (this includes the participants page).
* Activities supporting group overrides will not allow overrides for
groups that are hidden from all users.
* Activities supporting separate/visible groups modes will only allow
groups with the new "participation" flag enabled to be selected.
* Group messaging will be disabled for groups where members cannot see
each other, or cannot see the group at all.
The previous behaviour guessed the table filterset based on the table's
class name, but this is not very flexible if you wish to create an
abstract table class and use a generic filter for all implementations.
This change adds the ability to specify the filterset class, with the
default behaviour using the table class name as a prefix.
A new grunt task, upgradablelibs, has been added in order to get the
list of libraries that have a newer version in their repositories.
Co-author: Andrew Lyons <andrew@moodle.com>
If for any reason a Redis session lock is not being released, all subsequent
requests will wait to acquire the lock, forcing them to time out eventually.
This will happen till the original lock finally expires after the session timeout.
This sets the Redis session lock expiry time to whatever is lower,
either the PHP execution time `max_execution_time`, if the value was
defined in the `php.ini` or the globally configured `sessiontimeout`.
Setting it to the lower of the two will not make things worse it if the
execution timeout is longer than the session timeout.
For the PHP execution time, once the PHP execution time is over, we can
be sure that the lock is no longer actively held so that the lock can
expire safely. Although at `lib/classes/php_time_limit.php::raise(int)`,
Moodle can progressively increase the maximum PHP execution time, this
is limited to the `max_execution_time` value defined in the `php.ini`.
For the session timeout, we assume it is safe to consider the lock to
expire once the session itself expires.
If we unnecessarily hold the lock any longer, it blocks other session
requests.
Co-authored-by: Daniel Ziegenberg <daniel@ziegenberg.at>
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
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.
- Use span instead of dummy input elements because form elements need a
label
- Sufficient size for target is at least 24px by 24px (Success Criteria
2.5.8)
* Have the following classes extend qtype_essay_format_renderer_base
as they have been meant to:
- qtype_essay_format_editor_renderer
- qtype_essay_format_noinline_renderer
- qtype_essay_format_plain_renderer
* Add a question_display_options instance variable for
qtype_essay_format_renderer_base so its subclasses can use it to
generate the label for their respective answer fields.
* Add the question number to the answer fields if it's available.
* Improve multiple choice question accessibility:
- Label the multiple choice question groups appropriately by
enclosing them in fieldset tags and applying sr-only legend tags to
label them.
- Apply Bootstrap form-check classes to the radio buttons, so they
are rendered better and become responsive as well. This also helps
avoid the use of the table element for layout purposes when
rendering horizontal multiple choice sub-questions.
* Add an instance variable to question_display_options to store the
identifier associated with the question being rendered.
* This information can be used by question type plugins to improve the
accessibility of the answer fields being rendered by adding the
question identifier to the answer fields' labels.
* Adding the question identifier to the label can be achieved by using
question_display_options::add_question_identifier_to_label().
Co-authored-by: Tim Hunt <t.j.hunt@open.ac.uk>
- set the role of table to none
- moved aria-label from <table> element (that doesn't has role="none")
to the element that has the radiogroup role
- if it's not radiogroup/radio, it is list/listitem
- removed aria-label from the rubric table and used caption instead
Simple JSON reader which takes an issuer and reads its metadata. The
class is aware of request-centric rules, such as:
- The position of the well known suffix in the URL (per RFC8414)
- The requirement to have HTTPS auth server issuer URL
- The validity of query strings, paths and fragments in the auth server
issuer URL
,but makes no attempt to validate the config JSON returned.
This is very similar to MDL-77090, but at the time, I missed that this
also needed to be handled. (Question metadata is, I think, only used
by the quiz 'Try another question like this one' feature.)
Some behat scenarios were using labels with links to access some
specific PHP files.
Now that labels are also displayed in the course index, the region
needs to be defined in the step to access to the expected link
(otherwise, the one in the course index will be used).
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
When clicked, disable the "Edit mode" button just before doing a
redirection. It's been reported that after clicling several times the
button, an inconsistent state between the button and the editing mode is
achieved. The patch mitigates the issue.
Some extra options have been added to the activitychoosertabmode setting, to let
admins decide when to display the Recommended tab.
Apart from that, one of these values have be set as default value for this setting,
as suggested by the UX/PX teams. So the Starter and Full presets have been updated
too with the new values.
There is a Pull-request from Tim Hunt to the project in GitHub
(https://github.com/matthiasmullie/minify/issues/317) but, for now,
as it hasn't been applied yet, we need to manually cherry-pick
this change.
Enabling the course copy UI of MDL-64843 already without the capability 'moodle/course:view'
has the advantage, that a user with 'moodle/course:create', 'moodle/backup:backupcourse'
and 'moodle/restore:restorecourse' (typically a teacher)
will be able to use this UI for their own courses and with the additional benefit
of also being enrolled in the copied course.
Important: the role assigned to the user who copies a course is determined
by the value of the setting 'creatornewroleid', which defaults to 'editingteacher'.
* When the server is not reachable (or recordings are not reacheable)
the recording is set to RECORDING_STATUS_DISMISSED and is then
not visible in the list.
* Add a CLI script to solve potential issue with DISMISSED recordings
We now include more polyfills than we have ever done, and some of these
cause false-positives on our code to address modules which do not have a
name.
We started adding module names to modules in Moodle 3.8 at _build_ time
and kept support for modules which were not transpiled using Babel
before that point.
We no longer need to support this as all pre-3.8 Moodle versions are
long out of support. The most recent to go out of support was Moodle
3.5, which went completely out of support in May 2021.
We should not support code written and minified without a transpilation
phase from this time any longer.
When we unlock completion criteria and save changes 'viewed' criteria
still should be completed if user has viewed activity before. Seems
that that was broken by recent changes. Fixing here.
For hidden grade items we used to mark students as completed if
they have any grade. But this was not working correctly when we
also set pass grade for activity and completion criteria based
on pass grade. So we will have these completion states
Competion criteria 'Receive grade':
No grade - COMPLETION_INCOMPLETE
Grade visible, achieved passing grade - COMPLETION_COMPLETE_PASS
Grade visible, failed passing grade - COMPLETION_COMPLETE_FAIL
Grade hidden - COMPLETION_COMPLETE
Completion criteris 'Receive passing grade'
No grade - COMPLETION_INCOMPLETE
Grade visible, achieved passing grade - COMPLETION_COMPLETE_PASS
Grade visible, failed passing grade - COMPLETION_COMPLETE_FAIL
Grade hidden, achieved passing grade - COMPLETION_COMPLETE_PASS
Grade hidden, failed passing grade - COMPLETION_COMPLETE_FAIL_HIDDEN
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Co-authored-by: Tim Hunt <t.j.Hunt@open.ac.uk>
In PHP 8.2 and later, setting a value to an undeclared class property is
deprecated and emits a deprecation notice.
So we need to add missing class properties that still need to be declared.
Co-authored-by: Tim Hunt <t.j.Hunt@open.ac.uk>
Also:
* Fix some incorrect punctuation in two strings.
* Add a test to verify that this change does not cause a regression.
* Improve some PHPdoc comments.
Due to missing `AND k.script = 'rss'` condition, the query did not make
use of the existing script-value compound index defined for the table.
So it had to perform the full sequential scan for all rows when
searching for the token. This had serious performance issues on sites
with many users, especially in case on non-existing token / key.
Whenever some plugin is missing from disk but installed, the plugins
page (and the upgrade) shows them as "missing from disk". Still, the
code tries to do things with their directory that, now, is null.
That was silently defaulting to '' in previous php versions, but
php81 emits a deprecated warning. So we have to check for them.
Note: The original callback was incorrectly using the $PAGE output,
which cannot be relied upon in this callback. The best we can do here is
to add a notification to explain the situation.
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.
Unregistered sites will now display registration form to admins
after every core upgrade. Non core upgrades will not trigger the
form.
Registration form will not show for sites that are not set to
public.
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.
Since 0188af39 we've cached loaded reports, however there was an
edge case in those report sources that relied on the current user
as part of their own initialization (e.g. checking capabilities).
Co-authored-by: Marina Glancy <marina@moodle.com>
This is only randomly affecting to Moodle 4.2 with PostgreSQL 13
that, for some reason, returns the control to the script before
some files (.pid) have been deleted as part of the database
shutdown.
It seems that adding 1 second of sleep it's enough to workaround
that problem (there are examples of that in various Travis discussions).
So we are adding here 2 seconds, to ensure we are in the safe side
towards consistent passes. And we are applying the same strategy
also to the MySQL installation, just to keep them 100% parallel.
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.
Text elements must have sufficient colour contrast against the background,
restyling the link element when it is focused is to ensure colour contrast of at least 4.5:1
Following MDL-61880 you could turn on "Authenticate token requests via
HTTP headers" but not turn this off again. This change fixes this and
adds a Behat scenario to test toggling this checkbox is saved
correctly.
When a single template is reset, the string displayed in the modal should
not reference to "all templates" to avoid confusion, as it was shown in the
original prototype for the 4.1 database project.
Probably from historical reasons and because of layout of former Moodle
versions the container the feedback is situated in has 70% width and is
indented. In nowadays themes this looks weird and gives visual inconsistency.
Previously, it was possible to create drag-drop markers and onto image
questions without any drag items or drop zones. This was non-sensical,
and broke statistics calculations.
So, missing validation added, and random guess score calculation made
robust.
If you delete a file with a hash and then create another file with
the same hash, sometimes on EFS filesystems while trying to create
the new file, it returns true to the file_exists check even though
the file doesn't exist, but then fails other calls.
This change makes Moodle tolerate that behaviour.
When cache is disabled but temporary in-memory caches are allowed,
these were direct instances of cachestore_static with no loader, which
meant that the get_versioned and set_versioned functions did not work.
The reactive debug panel stops working when a new reactive instance is
created in the fly. This was not detected until the new dragf and drop
files into course uses a second reactive UI component in the same page.
A new developer setting that adds comments in the page HTML specifying
which template is used to render each part of the page. It can be set
form the site admin development tab and it is stored
as $CFG->debugtemplateinfo.
The get_items() and get_page_items() external methods should return
items only when the user has access. Otherwise, empty array for items
will be returned, with the exact error in the warnings parameter.
The component the audience belongs to is used by the cards exporter
for rendering the editing interface. Avoid duplication in determining
this value by defining appropriate method in the base class.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Since PHP 8.2, placing the dollar sign outside the curly brace is deprecated when
the expression inside the braces resolves to a variable or an expression.
Changes done to fix access order of attempts that need manual grading.
Changed button text from 'Save and go to next page' to 'Save and show next' to improve UI.
Move the construction of the useridlist $SESSION cache to when a key
is requested.
This stops the writing of $SESSION when backing up
or restoring mod_assign instances, which is neccessary since the
backup and restore scripts close the session when processing.
The session write_close was moved to the earliest point in the script
that does not modify the session. This is currently always after
$OUTPUT->header()
As part of a restore, the session is closed early so it does not
interrupt the users session during the restore.
Currently the restore controller rebuilds the course caches while
restoring. This inadvertently resets the navcache, which would edit the
session despite it being closed.
Because this tracker now adds logging for this behaviour, it means
restoring now outputs a debugging message as a warning.
To resolve the debugging message, the navcache is now reset just before
closing the session. This is allowed, since the caches are designed
to be volatile.
The script has been reorganised so that it closes the session as early
as possible, without causing session changes after close being logged.
In this case, after the table has been output it is OK to close the
session.
A snapshot of the session is now taken when write_close is called.
The session at shutdown is then compared to the snapshot. If changes
are detected, they are logged. This aids developers in seeing if
early session closes may be having unintended consequences.
Previously, newly added keys to the session were not detected. Objects
with the same properties were also incorrectly reported as different.
This commit improves this, and updates the unit tests to reflect the
new functionality.
In 4.0, the plugin name was added above their name. Based on the community
feedback and the research done by the UX team, this needs to be removed.
This commit removes the activity name when edit mode is disabled.
During request lifecycle ensure that column, filter and condition
persistents are only loaded a single time to prevent additional
database reads.
Provide invalidation method, used by the report helpers, to ensure
that the persistents are re-loaded appropriately.
Co-authored-by: Marina Glancy <marina@moodle.com>
Apart from applying the points described in readme_moodle.txt, the following
changes have been done too:
- The parameter $folderName from the method libraryToString() have been removed
and a new method, libraryToFolderName() has been added to the H5PCore API.
References to libraryToString() with the $folderName set to true have been
replaced to the new method.
- missing-main-library has been added and replaces in some cases to
missing-required-library.
- The framework saveLibraryData method must be called before saveLibrary
(h5p.classes.php file has been patched to leave the original order because
libraryid is required to save the itemid).
- The getLibraryId() method from H5PCore has been rewritten to use MUC, in
order to avoid PHPUnit failures.
Cache locking fails if the cache store supports multiple identifiers
(in core, the only two which do are cachestore_static and
cachestore_mongodb, so this is unlikely to cause severe problems).
The Preview questions icon shouldn't be displayed unless the user can
edit the feedback or access to the reports; otherwise, it's causing
confusion (especially when the feedback is not opened).
Replace the old course/dragdrop.js file (which was not even minimised)
to AMD modules and integrate them to the new reactive course editor.
From now on, a file can be drop over any course section, no matter if it
is in the course content or in the course index. It will also start
using the new process monitor to show the uploading state to the user.
In 4.0- version each time the course page is loaded the file handlers
are calculate din the backend and injected directly into JS using a json
encapsulation. With this new webservice the handlers can be obtained
directly from the frontend when needed.
This commit adds all the necessary CSS and logic to handle file dropping
into a reactive compoment. From now on, a reactive application can
handle both element drag&drop and file drop easily.
Create a new UI compoment to queue, execute and display errors on batch
processing. The first use of this component is when the teacher drops a
file into the course page.
Half of the times the normalise module is used is to get a single
element. However, because jQuery elements can contain multiple elements
the getList is always an array. Due to this in many ocasions we repeat
the getList(VAR)[0] line instead of having a more readable getFirst
method which only implies a couple of lines in the original code.
Basically, we only need to change:
- PHP 8.0
- MySQL 8.0
- PostgreSQL 13
Also, remove any php-xmlrpc installation because they aren't
needed since Moodle 4.1.
Finally, ensure that we aren't using mysql bin logs ever, because
they are huge and can make the 1Gb tmpfs to become full. This is
specifically needed for MySQL 8.0, because it comes with bin logs
enabled by default.
We have forked the mysql-action to achieve that:
https://github.com/moodlehq/mysql-action
Worth blaming travis because, after PG 11 and 12 using port 5433,
now they are back (in PG 13) to port 5432. Surely there is some
logic behind the undocumented ping-pong but...
Note that, despite the commit message, this is not possible. Moodle
3.11.0 (and 3.10.0) were developed in parallel with Moodle 4.0, and
when they were released, the master branch had already diverged, so
the master branch does not contain the comment lines:
"// Automatically generated Moodle v3.11.0 release upgrade line"
And they are needed to know which parts of the upgrade are safe to delete.
So we only proceed to delete all those steps from lib/db/upgrade.php
which version is known to be < 2021051708 (v3.11.8), master ones will
be, always bigger than that. We don't touch plugin upgrade scripts
because they may have different versions, not 100% matching the
2021051708 rule, so there will be an excess of steps there.
Note this is not a big problem, just a few more steps will be skipped and
that's all. Whenever we bump the Moodle requirements again (to Moodle
4.0 or up), the lines will be back and we'll be able to safely remove
all the steps before them. See previous Moodle requirements issues as
example.
The upgrade steps deleted by this commit weren't using any stuff from
upgradelib, tasks..., so there isn't anything else to be removed but
the core steps themselves.
Also includes an upgrade step to prevent upgrading from any
version < 2021051708 (v3.11.8) as anti-cheating measure.
Now that the required PHP version has been raised to php80, set
it in the composer.json file and regenerate everything, following
the instructions @ https://docs.moodle.org/dev/Composer .
The only changes, as agreed in the issue are:
- Raise min PHP version to 8.0 (from 7.4).
- Make the php-sodium extension required (was optional).
- Small behat/behat bump to 3.12.0 (from 3.11.0)
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>
Old-version backups (< 3.11.7, < 4.0.1) lose their course-level format
options when restoring into newer versions after the backup structure
was changed.
Also, course-level format options were not accurately restored against
the appropriate format type. And section-level format options were
redundantly stored at course-level and restored without mapping their
sectionids, thus creating lint.
The behaviour-specific data was getting corrupted when the regrade
recreated the first step, because $oldstep->get_behaviour_data() was
stripping off the leading '-' characters from the names, and they were
not being added back.
Some calls to the external_multiple_structure's constructor incorrectly
pass parameters where the value for the $description parameter is not
being passed. This results to invalid values being passed for the
$required parameter.
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.
Ensure column and filters provided by each source have sufficient
coverage. Re-factor individual entity/helper class tests to test
their own specific implementation only.
Remove now-defunct test fixtures.
The base time allows for the setting of the "current time" for the
purposes of the schedule test against recurrence configuration.
We should ensure the generator method allows this property so that
it can be used when calculating the initial "next send time" of the
schedule upon creation (and avoiding excessive recursion when the
actual current time is used for the calculation).
The TinyMCE menu has a significant issue with the Overflow style,
and the Boost theme heavily uses Overflow for drawer navigation.
Nest the dropdown menu container into the parent editor container makes it work correctly.
Co-authored-by: davewoloszyn <david.woloszyn@moodle.com>
Co-authored-by: xr0master <xr0master@gmail.com>
This commit will add missing fields as Time enrolled/started/reaggregated
in the filter/condition area for the completion entity when creating a course participants
as a report source in the Report Builder.
Calls to `str_ends_with` are polyfilled currently, as the method is
not present in PHP7.4 or earlier. We should avoid uses of it so as
to not trip up compatibility standard checkers.
quiz_settings::create() requires at least 3 DB queries, so the point is
that we can avoid doing that until after we have worked out if we need
to calculate statistics for this quiz.
Also, we order the list of quizzes to consier, to process the ones with
more recent attempts first.
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.
This commit will implement the Guzzle library in core
to make it usable in different locations.
Co-Authored-By: Andrew Nicols <andrew@nicols.co.uk>
Co-Authored-By: Safat Shahin <safat.shahin@moodle.com>
This commit will import the Guzzle library in
core. The imported libraries will be used in
the next commits to create an api for the http
client to be used in relevant locations.
Co-Authored-By: Andrew Nicols <andrew@nicols.co.uk>
Co-Authored-By: Safat Shahin <safat.shahin@moodle.com>
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.
This was used as part of the auth_config.php configuration system for
the auth subsystem, which was deprecated in Moodle 3.3.
It has never had any other uses in core.
Whilst it would be ideal to deprecate it by replacing with an
alternative, it would be impossible to test.
Changes in fbc2732d made the JS template string helper method
too strict in regards to what it considers a valid JSON object.
Co-authored-by: Mike Churchward <mike@brickfieldlabs.ie>
In PHP 8.1, any implicit conversion of float to int, which leads to a loss in precision, is now deprecated.
To avoid the error warning, the code must explicitly convert a floating value to an integer.
Pass correct parameter/type in field constructor (`XMLDB_NOTNULL`)
for consistency. This corrects the definition and preserves the
truthyness of the value that was incorrectly passed previously.
This commit makes the following changes:
* convert to use ES6 module format
* migrate from a jQuery Deferred to a native ES6 promise
* add a native-Promise-like instantiation
New uses are documented in appropriate jsdoc with examples.
Whenever the page_requirements_manager::js_fix_url()
is called with null url, it must throw an exception and emit 0 warnings.
It's covered by an explicit test:
test_js_fix_url_coding_exception with data set "Provide a null argument"
The json_decode function does not accept a null, which is the
traditional default for get_user_preferences. By passing a default of
am empty string we avoid issues in PHP 8.1.
Revert the lang string "Add an activity or resource before [xx]" back to
its original form of "Insert an activity or resource before [xx]" in
order to avoid conflicts with the "Add an activity or resource" button
especially when running Behat tests.
A new signature was introduced to ImagePolygon functions from PHP 8.0
with the previous signature deprecated in 8.1.
For PHP versions 8.0 and later we should use the new signature. Once PHP
8.0 is the required minimum version, the legacy version should be
removed.
This report source differs from the original badges source (5274ee5a)
by making `user` the primary table of the report, allowing for reports
on all users regardless of whether they have been awarded badges.
Added a few functions to remove the toolbar button, menubar and sub-menu items.
One of the implementations is to remove the justify alignment in the toolbar and
the sub-menu items to aid the accessibility aspect of the TinyMCE editor.
This change makes the regrade progress bar send frequent updates to
the web browser, even when processing within the same grade item, to
avoid timeout problems when there are a large number of users.
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!
With Moodle 4.1 and up, we have stopped to work with strftime(),
because it's deprecated for PHP 8.1 and up and, instead we are
using core_date::strftime() that performs the conversions using
the php-intl extension.
The original strftime() had a lot of OS-specific dependencies,
both about locale names and charsets used to provide the information.
Instead, the php-intl gets rid of all those OS-specific nightmares
and only uses ICU/Unicode locales and UTF-8.
Because of that, the old utf-8 => windows charset => utf-8
conversions aren't needed anymore. Hence, removing them.
MDL-75596 added an "Insert an activity or resource before ..." button
that allows adding an activity or a resource before or after another
activity in a course. This renders the button in the DOM (although
hidden and shown only on mouseover) when there's at least one activity
in the course.
In this issue, the button's label was renamed to "Add an activity or
resource before ..." and it now obscures the normal "Add an activity or
resource" button at the bottom of the course section from being
interacted on when running Behat tests.
This failure happens in the create_activity.feature file as it creates
two DB activities in a single scenario – one when "usecomments" is
turned on and one when it's turned off. Given this limitation, a
workaround is to split the "Teacher cannot enable comments if they are
disabled at site level" scenario into two scenarios with the first one
testing when "usecomments" is on while the second scenario testing when
"usecomments" is turned off.
We created a list of placeholder selectors in placeholderSelectors option.
The purpose of this list is to indicate the contents that are only shown
in the editor and not to the users, by that way, we can decide to apply or
not to apply rules, styles, etc... to these contents
* Add "Preview" and "Use this preset" menus to the preset plugin page's
burger menu
* Refactor importmapping dialogue to use the 'data-action' selector
* Add a new set_kebab_trigger action menu method
* Use set_additional_classes method for action menu
* Remove redundant parameter to add_action_menu
PHP before version 8.1 automatically converted to int if the function
parameter (or array key) is expected to be int. PHP 8.1 shows notice in
this case
PHP before version 8.1 automatically converted stdClass or 'false' to arrays if
function parameter expects array (for example, "reset").
PHP 8.1 shows notices in these situations
Currently, the unordered list in the course content follows the original style and
is ordered from the top of the content section, which is not part of the description.
That's why the first occurrence of the list is the 2nd level which uses the circle as the marker/list type
instead of the disc. In contrast, the editor always uses the original style.
so we need to reset the CSS of the unordered list in the description to the original
which the disc as the first marker, a circle for the second marker,
and a square for the following levels.
Allow users to see their own private messages even if they don't have
mod/forum:readprivatereplies capability
Add behat test to avoid MDL-67109 issue.
Co-authored-by: Martin Putzlocher <mputzi@users.noreply.github.com>
At the moment, quiz_statistics\task\recalculate gives no useful
information about what it is doing, which makes it hard to investigate
if the task fails. This commit makes it more usefully verbose.
Also, following this change, one instance of this task will not
run for more than one hour at a time.
As part of this commit, I have added a new helper mtrace_exception.
to consistently log exceptions in scheduled tasks. It is sad to
add a new function to moodlelib.php, but that seemed the logical place.
Looking at other tasks, this method is badly needed. Many are just
logging the ->getMessage() which is normaly insufficient for proper
debugging. However, swiching all existing tasks to use the new function
will need to wait for a future MDL.
As in MDL-74691, we need either or both of these fields, meaning either
one could be omitted. This just supports that as per the fix made in
MDL-74691.
This covers the following cases:
1. Where only the 'lineitem' service endpoint is provided
2. Where only the 'lineitems' service endpoint is provided.
Existing tests already cover the case where both are provided.
After the user creation, the system must call an update function to update profile_fields_*.
We also provided two functions into user/profile/lib.php to get available from other areas.
We added PHP unit testing for new public functions and
the Behat tests for custom profile fields with locked and unlocked statuses.
Co-authored-by: Matt Porritt <matt.porritt@moodle.com>
Since the <ul> and <li> elements for listing the quiz questions and
elements are mainly used for layout purposes, it is better to set the
role of these elements to presentation only.
The aria-label for the <li> element is inappropriate and can also cause
confusion to screen reader users. Better to remove this. Though this is
being used in the confirmation modal so better to just change it to a
data attribute.
Since the <ul> and <li> elements for listing the quiz questions and
elements are mainly used for layout purposes, it is better to set the
role of these elements to presentation only.
Add a default sr-only section heading in order to prevent having an
empty heading for the quiz section which is not good for accessibility.
Plus:
- There's also an existing issue where filters are never applied to
section headings on load. This has been fixed in this commit as well.
- Button role added for the edit section button which enables it to
take advantage of the theme highlighting when it receives focus.
I thought about renaming the class to just plain attempt, but I
acutally think quiz_attempt makes it clearer what this is. Also
not changing the name massively reduces the pain for everyone
updating their code (including me right now!)
Code should just directly use moodle_execption.
Note, some of the existing uses (in web services) were passing a
literal message, not a language string identifier, but I decided
now was not the time to try to fix that.
Update oauth2 to allow mapping of provider attributes against
user profile fields. Fields can also be locked to prevent
user changes.
Co-Authored-By: Michael Milette <michael.milette@tngconsulting.ca>
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>
- Markdown is saved to the DB, no conversion to HTML to keep
ability to reedit the content without beaking it.
- Blockquote element is styled that it's distingishable from
normal text.
See also f4fb91ae. PHP8 now supports named arguments in calls to
the `call_user_func_array` method, which can result in differing
behaviour compared to PHP7, e.g. in this case it was throwing an
exception due to an unknown named parameter.
Displaying the course card menu only on hover is bad in terms of
accessibility as it prevents keyboard-only users from accessing the
menu. So it's better to always display them.
If a user drags an activity and press the ALT key, the activity will be
duplicated in the new place instead of simply moved. This behaviour
is quite normal in many softwares and it is a non intrusive usability
improvement.
Adding an activity with a grade item previously triggered a course
regrade, which can be extremely slow. This change makes it so that on
courses where regrading is slow, when you add an activity that
requires regrading, it takes you to an intermediate screen with a
progress bar.
This change also fixes a bug in the existing grade overview report,
which shows grades across multiple courses, but previously only
checked, and if necessary regraded, one course.
The courseindex init methods contains static references to the class
name instead of a dynamic "this" reference. This makes harder for format
plugins to extend those classes as they need to override the init method
for no specific reason.
I have opted not to go through any deprecation process for this as there
is not really much we can do about it anyway. There is no loss of
functionality - just a change of where things are served from.
The Yahoo! CDN was only used for http connections as it did not
officially support SSL, and it is not guaranteed to continue existing in
the future since YUI has been deprecated for a number of years now.
Including in this commit:
- Prevent the tour to be displayed if the CSS is disabled by the browser
- Removed redundant tabindex=0 attribute on the target
- Created extra code to re-calculate the suitable position of the tour
- Fixed tour issue in 200% and 400% zoom
- Fixed tour highlight is not correct in 200% and 400%
The highlight and unhighlight sections applies only to the format_topics
plugins. With this commit all the related code is migrated to reactive
component and format_topics extend the current course editor to add all
the logic
Add 'versionfromdb' to opt-in for VERSION() query.
mariadb_native_moodle_database::get_server_info() has been removed
to avoid the specific override: everything is now managed in the base
method which takes care of both MySQL and MariaDB versioning details.
New protected methods have been added too to improve the testing strategy,
allowing to mock the results from an actual DB connection.
The new testing strategy enables simulating even the Azure scenarios.
When the param string contains a left curly bracket as the first character,
the system will assume the string is a JSON string and will be parsed and returned as an object.
But in some cases, the string is not JSON and will return an error if the system parses it.
For example, a user might have used the course name with a left curly bracket as the first character.
Adding a double quote after the left curly bracket to differentiate between string and JSON string,
so it can be safe to parse the string, and also added try..catch to ensure that the parsing creates an object type.
The new library version should work without problems with older
php-mongodb extension versions:
- v1.15.0 (paired version, tested).
- v1.14.1 (tested)
But, in order to keep them paired, the requirements have been
raised to the extension v1.15.0. See:
https://www.mongodb.com/docs/drivers/php/#compatibility
This updates the library from v1.13.1. Fixes some php81 warnings
leading to failed/risky tests.
The new library version should work without problems with older
php-mongodb extension versions:
- v1.15.0 (paired version, tested).
- v1.14.1 (tested)
During install/upgrade, caching is disabled. This change provides a
way to temporarily enable caching (using in-memory cache storage only)
within a specific function; caches are deleted afterwards.
Adding this to two locations improves install performance quite a lot.
Caching is not enabled during the parts of plugin installation that
can vary for individual plugins (install.php/upgrade.php) as these
might be relying on its absence, for example by making direct database
changes.
Adding new capabilities can be slow if they have permissions cloned
from an existing capability, and if there are many places in the
database where the existing capability was overridden.
This change optimises the clone-capability performance, approximately
halving the number of queries required.
Adding multiple capabilities was very slow because it clears and
rebuilds the capability cache after each one. This change makes it
clear the cache once after adding all capabilities, before adding
permissions to the archetype roles.
Webservice doesn't need to set the page URL. Hence,
any function that requires a page URL will raise debugging output.
The patch fixes it by adding a fake URL for the webservice.
The patch also removed assertDebuggingCalled because it is no longer needed.
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.
jQuery submit events do not trigger the native submit event, and
therefore the TinyMCE submit listener is not called. This means that the
tinymce.triggerSave() function is not called, and the editor content is
not saved.
To work around this, if there is a form for the Node that the editor is
attached to, the jQuery submit event is listened to and the save event
called.
Unfortunately it is not possible to use a global jQuery submit listener
added to the document to save all editors on the page.
In some cases, this is too late because the bubbling process means that
a child node consumer may attempt to consume the data before the
delegated handler is called for the document.
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
If the default dashboard contained a block that was missing from disk
then "Exception - Call to a member function instance_copy() on bool"
was shown to users. This change only shows an error with debugging
enabled which now includes the name of the missing block.
Use the full result from download_file_content within the component
installer to ensure that spurious debugging is not emitted during
operation. Handle validation of the returned data consistently.
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 part of the code in 'question_category_object.php' 'update_category' method was used before Moodle 4.0 version for renaming
the random questions in an updated category. For Moodle 4.0 and later versions, it is unnecessary as the details of
random questions are no more stored in 'question' table but in 'question_set_references' table.
The method call move_question_set_references handles the same.
AMOS BEGIN
MOV [config_recordings_sortorder,mod_bigbluebuttonbn],[config_recordings_asc_sort,mod_bigbluebuttonbn]
MOV [config_recordings_sortorder_description,mod_bigbluebuttonbn],[config_recordings_asc_sort_description,mod_bigbluebuttonbn]
AMOS END
We will not calculate old maker position again and using old data.
This will make sure the position of makers is correct
when the background image is smaller than dropzone.
Right now we have the information only in docs:
- https://docs.moodle.org/dev/Core_APIs
- https://moodledev.io/docs/apis
And, in fact, we are crawling those pages to get the information
from various tools (moodlecheck, CiBoT...). Obviously, that's far
from ideal, the source only has the current list of APIs, and
there isn't much information there but the names.
So we are moving the source of information to be in core, so it
can be modified between branches, and contains richer information:
- The component the API belongs to, usually a subsystem or core.
- If the API can be used as level 2 namespace.
- If the API can be used as level 2 namespace out from its component.
Note that all that information has NO USES right now in core (and maybe
never will), but tools/checkers will benefit enormously by having that
information at hand, so we can check for namespaces, categories and
other bits way better.
Also, once we have this, the APIs dev documents linked above, surely
can be improved by being automatically generated and include all the
meta-information available.
It also includes a very basic json schema validating the basis. It can
be tried online @ https://www.jsonschemavalidator.net , or any other
tool. PHP requires extra libraries to be able to perform the validation.
Covered with unit tests, both api-related functions and structure validation.
Remove negative margin from the "clear my choice" anchor.
With the negative margin, when the containing box height is calculated
the anchor's hidden style overflows beyond the boundary and triggers
the overflow:auto to add a scroll bar.
... or anything else with length = 0. This got broken in MDL-71696.
The only way to fix this kind-of involves and API change to
quiz_report_get_significant_questions. However, it is only changing
the external API of this function back to how it was before the 4.0 release,
and the chnages in 4.0 were never documented, nor, I would guess, intended,
since they just broke things.
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>
When the site uses custom user text fields that would show up in a
user list, the filter for firstname or lastname didnt work anymore.
If a filter was used, no result would show up. Users where shown only
when the course reports where used without filters.
It was detected that 'This month' option may cause failures in builds in between months
so using the 'Recent and next 60 days' still fixes the weekend scenario and also the
in between months issue.
When the keyword participant filter was used with multiple values
in conjunction with other participants filters, the boolean logic
was effectively:
X AND Y OR Z
When what we actually wanted was:
X AND (Y OR Z)
Set type of the report paging setting to integer, to ensure usage
of it is predictable. Unsupported operated type errors were thrown
on PHP8.0 when it's value contained a string or was empty.
This adds a new method to the assignfeedback edit pdf library
to specify user data file areas that will return just the meaningful
annotated feedback pdf.
get_file_areas has been updated for this plugin to return all
file areas related to assignfeedback_editpdf, and should stop
producing orphaned files and records when a course reset is done.
Thanks to @toniginard who provided a base solution for me to work
off.
The method already loaded the validated persistent model data on the
previous line, there's no need to do it again (while also trying to
load unvalidated properties).
The assumption that the settings checkbox always stores boolean values
(e.g. 0/1) isn't correct. The `perfdebug` configuration instead uses
the values 7/15.
Default value of the $flag argument changed in PHP 8.1 from ENT_COMPAT to ENT_QUOTES | ENT_SUBSTITUTE
To ensure consistent behavior across different PHP version the second parameter is now required for the functions:
htmlspecialchars(), htmlentities(), htmlspecialchars_decode(), html_entity_decode() and get_html_translation_table()
Including in this commit:
- Convert existing Behat scenario to use Data generators
- Convert existing Behat scenario to use Provider
- Add new scenario for Include users setting in General restore page
This changes a couple of cases, in postgres and oracle where we
were trying to calculate strlen() on null values (that causes a
PHP 8.1 warning).
Also, at the same time, fixing another case in xmldb_field, it
was detected that it had a bug around precision (being set to the
value of type), being used as object property, when it's not, so
it was also fixed by moving the code logic a little bit.
Note that the bug has been there since inception because there
isn't any code in Moodle using that ->precision property. It
was just detected thanks to PHP 8.2, but that's another story. :-)
Verified that with the patch:
- All DBs can be installed (phpunit install)
- Both DDL and DML pass without any ddl/dml warning (in all DBs).
Prior to this change, all the line endings in the imported HTMLPurifier
library were using CRLF (\r\n aka Windows style), but the HTMLPurifier
source and also the downloadable artefacts use LF (\n aka Linux style)
as line endings. This has been the case since
510d190382 when with the commit
"MDL-38672 import HTML Purifier 4.5.0" all line endings were changed
from LF to CRLF. There was no comment in the commit on why this change
was done.
As the original source uses LF, this commit partly reverts
510d190382 and goes back to LF as line
endings.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
Set needsupdate to finish to avoid regrading each call.
Add error messages for regrading.
Display the error message on the quiz page.
Co-authored-by: Mark Johnson <mark.johnson@catalyst-eu.net>
In some cases, we get an error message such as:
Wrong parameter count for Redis::zRem()
Within the delete_many function. This function requires at least one
key to be supplied, but if delete_many is called with an empty array,
we will call it with no keys.
Fixed problems in lateral navigation and grading when
a SCORM has more than three levels
The SCORM module has problems with full screen display,
completion tracking and grading in SCORM.
Also resolves issues MDL-74992, MDL-73249, MDL-75407
2022-10-05 12:40:06 +02:00
10716 changed files with 397461 additions and 340038 deletions
{"version":3,"file":"block_management_table.min.js","sources":["../src/block_management_table.js"],"sourcesContent":["// This file is part of Moodle - http://moodle.org/\n//\n// Moodle is free software: you can redistribute it and/or modify\n// it under the terms of the GNU General Public License as published by\n// the Free Software Foundation, either version 3 of the License, or\n// (at your option) any later version.\n//\n// Moodle is distributed in the hope that it will be useful,\n// but WITHOUT ANY WARRANTY; without even the implied warranty of\n// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n// GNU General Public License for more details.\n//\n// You should have received a copy of the GNU General Public License\n// along with Moodle. If not, see <http://www.gnu.org/licenses/>.\n\nimport PluginManagementTable from './plugin_management_table';\nimport {refreshTableContent} from 'core_table/dynamic';\nimport {call as fetchMany} from 'core/ajax';\nimport Pending from 'core/pending';\nimport {fetchNotifications} from 'core/notification';\n\nexport default class extends PluginManagementTable {\n constructor() {\n super();\n this.addClickHandler(this.handleBlockProtectToggle);\n }\n\n /**\n * Set the block protection state.\n *\n * @param {string} plugin\n * @param {number} state\n * @returns {Promise}\n */\n setBlockProtectState(plugin, state) {\n return fetchMany([{\n methodname: 'core_admin_set_block_protection',\n args: {\n plugin,\n state,\n },\n }])[0];\n }\n\n /**\n * Handle toggling of block protection.\n *\n * @param {HTMLElement} tableRoot\n * @param {Event} e\n */\n async handleBlockProtectToggle(tableRoot, e) {\n const stateToggle = e.target.closest('[data-action=\"toggleprotectstate\"]');\n if (stateToggle) {\n e.preventDefault();\n const pendingPromise = new Pending('core_table/dynamic:processAction');\n\n await this.setBlockProtectState(\n stateToggle.dataset.plugin,\n stateToggle.dataset.targetState === '1' ? 1 : 0\n );\n\n const [updatedRoot] = await Promise.all([\n refreshTableContent(tableRoot),\n fetchNotifications(),\n ]);\n\n // Refocus on the link that as pressed in the first place.\n updatedRoot.querySelector(`[data-action=\"toggleprotectstate\"][data-plugin=\"${stateToggle.dataset.plugin}\"]`).focus();\n pendingPromise.resolve();\n }\n }\n}\n"],"names":["PluginManagementTable","constructor","addClickHandler","this","handleBlockProtectToggle","setBlockProtectState","plugin","state","methodname","args","tableRoot","e","stateToggle","target","closest","preventDefault","pendingPromise","Pending","dataset","targetState","updatedRoot","Promise","all","querySelector","focus","resolve"],"mappings":"uhBAqB6BA,iCACzBC,2BAESC,gBAAgBC,KAAKC,0BAU9BC,qBAAqBC,OAAQC,cAClB,cAAU,CAAC,CACdC,WAAY,kCACZC,KAAM,CACFH,OAAAA,OACAC,MAAAA,UAEJ,kCASuBG,UAAWC,SAChCC,YAAcD,EAAEE,OAAOC,QAAQ,yCACjCF,YAAa,CACbD,EAAEI,uBACIC,eAAiB,IAAIC,iBAAQ,0CAE7Bd,KAAKE,qBACPO,YAAYM,QAAQZ,OACgB,MAApCM,YAAYM,QAAQC,YAAsB,EAAI,SAG3CC,mBAAqBC,QAAQC,IAAI,EACpC,gCAAoBZ,YACpB,wCAIJU,YAAYG,wEAAiEX,YAAYM,QAAQZ,cAAYkB,QAC7GR,eAAeS"}
// "Profiling" settingpage (conditionally if the 'xhprof' extension is available only).
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