Every backup and restore controller always need to call
to their ->destroy() method in charge of finalizing loggers,
closing file handlers and destroy some circular references.
This is specially important within unit tests, because
open resources cannot be reset between tests, causing problems,
specially within Windows that locks all those files.
This patch adds better core support for copying courses.
There is now a simplified and dedicated UI for copying
courses. This can be accessed from the course context
menu or course management screens.
All backups are done asynchronously and there can be multiple
copies of a course in flight at once.
This change enables the gherkinlint rule to require a new line at the
end of the file. This change is in keeping with existing Moodle coding
style guidelines.
* refactor run automated backup
* new backup status of Queued
* adhoc task for backing up a course
* course locking for course backup adhoc task
* use Moodle locking instead of custom locking for run automated backup task
* add unit tests to extracted methods
Fix condition where if an asynchronous backup
completes via the UI and the user refreshes the
page a new backup was created with a mismatched
backup id.
This caused a permanently failing ad-hoc task
and blocked further backups for that course.
This lets plugins other than activity modules backup and restore
question attempt data.
The old backup_questions_activity_structure_step and
restore_questions_activity_structure_step base classes still exist
and work exactly the same way they did before (because they use the
trait) so this change is completely backwards compatible.
To make this work fully, a few other things in the code had to be tweaked:
* Adding restore path elements had to consider the possibility of grouped
parents in more places.
* I needed to add protected get_task() to the restore_plugin class.
I don't think that is a problem.
* In the restore trait, the process_question_... methods needed to be
changed to public for some reasons to do with PHP traits that I don't
fully understand. However, I don't think this change is a problem.
* The way question_usage restore got the new contextid had to be changed
(or it did not work in activity contexts), but the new code looks like
a better way to do it anyway so that is good.
Allow both UI and automated backups to be created without
including files. Instead include only file references.
This is essentially implementing "SAMESITE" to backup files
instead of only for import and export functionality.
A new backup setting to include files (defaults to yes)
has been included.
The restore process will also look for and attempt to
restore files from the trashdir as part of restoring
backups. Additionally to support this process the
ammount of time files are kept in trashdir before they
are cleaned up via cron is also adjustable via admin
setting.
