I looked at randmly at some tasks from last year:
Some were given 4h, 24h, 48h or 96h. I guess that tasks of 24h would be
a good estimate to plan for 24h (without assuming they will be
programming non-stop, of course :).
Some tasks were quite open, such as "Find and report a bug in $product"
or "Find a security vulnerability (a file that crashes the program)".
I'm not sure if asking for something we don't know if it's doable is
appropiate, though. OTOH they allow multiple instances of the same task.
Others like update $PROGRAM to use this new API doesn't seem to have
been successful. I don't think tasks of that level are appropiate (we
could place a few, but without an expectation of being solved by the
A good thing is that I don't think that reviewing a task will take
longer than 5-10 minutes. So it shouldn't be a burden on the mentors.
Seems a workflow like the one used on OTRS. The key is having a big pool
of developers so a few ones don't need to review all of them.
An easy way to get articles would be in the documentation front, asking
for a couple of wiki pages documenting something, a tutorial (with
screenshots) on installing MediaWiki... Creating X new translations for
MediaWiki or its extensions on translatewiki would also be an easy way
of producing tasks.