BugTask.conjoined_master shouldn't use shortlist()

On Sun, May 04, 2008 at 03:56:03PM -0000, Graham Binns wrote:
> shortlist() is used in BugTask.conjoined_master to listify the BugTask's
> Bug's BugTasks sequence. If this warning is getting raised this often it
> should probably be changed to use list().

No, it shouldn't use list(); shortlist() is the right thing to use here,
since whenever getting an unknown (unbounded) number of things from the
db, we want to put a limit on the number of rows returned.

The right thing to do there is to increase the limit. You can specify
the limit when calling shortlist().

The problem attributes in the interface can be annotated with the nosnapshot directive

Lp has about 100 pathoglogical bugs. I suspect we need to increase the size of the shortlist, of allow users to delete invalid bug tasks that could not be retargeted.
 # bugs | # bug tasks
--------+-------
     16 | 15
     17 | 11
     18 | 19
     19 | 12
     20 | 6
     21 | 13
     22 | 4
     23 | 5
     24 | 8
     25 | 3
     26 | 3
     27 | 3
     28 | 3
     30 | 5
     32 | 1
     34 | 1
     35 | 3
     36 | 1
     37 | 1
     40 | 1
     44 | 1
     45 | 1
     46 | 4
     47 | 1
     52 | 3
     54 | 2
     71 | 1
     72 | 1
     73 | 1
     75 | 1
     92 | 1
   159 | 1

I think we should pragmatically just raise the figure for now to (say) 200, finish the bug task:+index performance work (potentially including deletes) and then reevaluate.

Fixed in stable r12530 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/stable/revision/12530>.

Fixed in db-stable r10261 (http://bazaar.launchpad.net/~launchpad-pqm/launchpad/db-stable/revision/10261) by a commit, but not testable.