Now we are checking for the need to execute the allocator every time the
main view.php is displayed to any user. Previously this was done by cron
only.
Removing the table "scheduled" created by the XMLDB by default (it's a
bug in XMLDB, I'm just too lazy now to fix it). Adding new fields to
hold the result of the most recent execution.
The idea is to build a plugin that will internally call the random
allocator when the submissions deadline is reached. The teacher is able
to pre-define the configuration for the random allocator.
The random allocator's init() method was split into two logical
parts. The init() now handles data from the form and executes the new
execute() method to do the actual job. The execute() method will be
called by the future scheduled allocator, too.
When generating the output of the allocation results, the names of
reviewers and authors are obtained based on the current state of assigned
roles and the 'musthavesubmission' setting. Therefore, when we want to
display information about a re-used allocation, the name of the author
or the name of the author may not be available.
In case of reviewers, we first try to reload the list including the
users without own submission. If it does not help, we just display the
id of the user instead of the name. In case of authors, we just display
the id the user. I can't see much point of eventual fetching the user
record from the database just to populate the list that almost no-one
reads anyway.
The protected method get_allocations_recordset() is a relict from
earlier development phases of Workshop module. Removing it as it
confuses recent Eloy's recordset usage detection tool.
There was a bug as these users were ignored only as squares but not as
circles. This patch makes sure that no new allocation are added to
not-grouped users in visible group mode. However, the options 'Remove current
allocations' and 'Add self-assessments' apply to them intentionally.
Also, the user is warned if there are some not-grouped users found in a
workshop in visible groups mode or separate groups mode.
The patch wraps that foreach ($circles as $circleid => $circle)
loop by yet another one for() loop. Reviews are allocated iteratively
now. During the first iteration, we try to make sure that at least one
circle link exists. During the second iteration, we try to allocate two,
etc. Circles are shuffled at the beginning of each iteration.
This is supposed to improve the randomness of the allocation.
The patch also fixes shuffle_assoc() implementation. The previous
implementation actually did not work at all. Also, that removed called
to shuffle_assoc() was redundant here.
