Comment 9 for bug 618849

I re-checked the test_attachments.py script I did for reporting bug #424671, and measured about a 35-40% reduction in run time for the particular case I'm interested in. However, the script terminates with an assertion so I can't be certain the performance improvement is real or just that the script exits early... I'll optimistically assume the former. :-) Details can be viewed on that bug report.

For timeout/oops with attachments, bug #317094 has been my test case (reported as launchpad bug #583553); I examined this with several folk at the epic without success. Unfortunately it still times out:

https://bugs.edge.launchpad.net/ubuntu/+source/xf86-input-evtouch/+bug/317094
https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1713EB2411

However, despite this problem, it appears the time has improved a good deal:

WAS:
SQL time: 5202 ms
Non-sql time: 15510 ms
Total time: 20712 ms
Statement Count: 506

IS:
SQL time: 6168 ms
Non-sql time: 7378 ms
Total time: 13546 ms
Statement Count: 413

Anyway, I don't know how accurate or relevant all this data is, but it suggests to me things are getting measurably better, although still generally broken.