still producing too much data and overflowing PHP memory on busier
sites (eg moodle.org).
Now, there are more database queries, which is unfortunate, but the
data is much more specific, and no sorting needs to be done, so
this is a performance boost.
I don't know how these will cancel out ... my guess is that very
small sites may be very slightly slower on the course page, but
that large sites will be much faster.
Let's see.
Moved COURSE_TEACHER_COLOR out to style sheets where it belongs (.teacheronly)
Added some efficiency when printing recent activity
- don't print more than one weeks worth of updates
- don't keep re-parsing logs once they've been used
same order. Note that shuffled ANSWERS may still appear reshuffled.
When viewing reports, all shuffling is switched off so that quizzes
always appear in the default order (for easier comparison).
Also a few tweaks, robustness etc
When you add a random question to a quiz, then a question
is chosen randomly from the same category.
Questions are not repeated in a particular quiz attempt.
This still needs more testing, but it's looking OK so far.
created. Works OK after limited testing.
I've also renamed some strings to do with "Random Match", so that this
question type is now called "Random Short-Answer Match".
Later there will be a new 'Random Match' which randomly selects one of the
existing "Match" questions.
... this grade is used as the default grade when adding questions
to a quiz.
At the moment it's just so that some imported questions can have
different defaults, but a field should be added to all question
editing pages.
of questions. These use 2 or more short answer questions at
random to construct a questions where you have to match
answers to questions. Only lightly tested so far.
Quiz questions can now be edited with Richtext editor.
Sundry little fixes along the way.
Moodle tables.
ie user -> userid in many tables, plus in user_students
start -> starttime and end -> endtime
I've just done all this as carefully as I could ... I don't think
I missed anything but it's pretty intensive work and I'd be fooling myself
if I didn't think I'd missed a couple.
Note that this version should pretty much be able to bootstrap itself
using PostgreSQL now ... but this is untested
