All the page headers work correctly also on the site course.
On the site page the modules don't require login unless necessary or required by $CFG->forcelogin.
1) get_course_students, get_course_users and count_course_students when called with course=site will now use get_site_users instead of get_users. This I believe was the consensus reached in the discussions of how front-page activities should behave.
2) all functions can handle a list of exceptions now.
3) get_site_users now returns users in the order admins, teachers, students. Similarly for get_course_users. This makes the sorting bug 1727 a bit more bearable
4) new function search_users
The whole thing is really a mess because each function has slightly different conventions for its arguments. But the beta is too close to tidy this up now.
Upon login, the last access to each course is stored in the USER session
variable.
The earliest date of recent activity now shown is now the earliest of:
- this last access time
- the last login time
If this value is older than two days, then just use two days instead.
Modified get_course_students and count_course_students to handle a request for course 0 (will get all users, even if they are in no other course).
Changes done to for admin/user.php, should be compatible with all previous function calls.
1. also return confirmed field from user table and
2. include users in dataset who have not confirmed
(Used to display unconfirmed users in admin/user.php)
Logs now include a field called modid which contains the coursemodule id.
This makes it now possible to
- see complete logs per-activity
- do backup/restore of logs
The upgrade process will currently try to scan all the old logs and
rebuild this field based on available data (especially forums).
STILL TO DO: alter all the non-forum modules to send the coursemodule id
that returned id-column was string and some modules were thingking that theres
insert error if return value is string.
Right now all id-colums is always integer so
this should be ok.
Petri Asikainen rewrote insert_record to use ADOdb functions. It might
be very slightly slower but it gets rid of all the horrible not-quite-working
PostgreSQL workarounds the old one contained.
It worked for him on PostgreSQL 7.4 and for me on MySQL 3.23 and 4.0.15,
so I'm checking it in.
Please test it thoroughly on your test systems.
Since it writes data it has the potential to stuff things up, so be
careful on production systems for a few days.
