Before this patch it was possible for the student displayed on the grading page to
not be the student that the user selected to grade. This would occur if:
1) The user had the table ordered by a value that could be modified,
for example Last modified (submission), Grade, Last modified (grade)
2) Another user performed an action that was recorded in Moodle in the time
between the user generating the table and clicking on a grade link.
If a user did not notice a different user had been loaded it could result in them giving
a grade to the incorrect user.
This patch ensures that the state of the table is cached every time it is viewed by a user
who has the capability to grade.
The previous 2.9 implementation of is_web_crawler() used stripos() in
certain cases. The unit tests re-added in the previous commit revealed
that certain crawlers (such as BaiDuSpider) were not correctly detected
in the new refactored implementation.
It seems lesser evil and safe enough to use /i in the regex search even
though it is not 100% same logic as before - as stripos() was used in
some cases only, not always.
In MDL-50891, the is_web_crawler() was refactored into a core_useragent
method and the function itself was deprecated. However, there were no
unit tests kept to check the backwards compatible behaviour. It turned
out that the deprecated function leads to PHP fatal error due to a typo.
This patch fixes the typo and brings back the previous unit tests. To be
able to explicitly check the raised debugging message, the test case
now must be subclass of advanced_testcase.
Additionally fixes missing info about the function being deprecated.
The 'completion enabled' flag does not reflect any course data, its a
course setting. Previously when doing a backup deleteting existing
course content this was forced off when it should not have been.
The problem with the previous implementation was that the table's
$this->prefs can contain valid non-empty value in its default state -
the default column to sort by. On resetting, we must not throw away
these default prefs.
This patch simplifies the TABLE_VAR_RESET interpretation. If such an
HTTP parameter is passed via the request, the table simply behaves as if
there were no previously stored preferences (does not matter if coming
from the current session, or from the persistent cross-session storage).
The logic that decides on whether or not the reset widget should be
displayed is put into a new method can_be_reset() with unit tests
attached.
Finally, the previously private method render_reset_button() is now
protected and the reset widget is given a new semantic CSS class.
