Comment 10 for bug 424671

Revision history for this message
Bryce Harrington (bryce) wrote :

In light of bug #618849 being fixed, I re-ran this script for case #6. Originally I'd measured a time of 1:18; the script now completes with a time of 0:49.29.

This is being run on the same hardware, with the same version of Ubuntu as the original test. The bug report under measurement in this test case (bug #259156 ) has received a few comments but no additional attachments. So while the testing conditions aren't exact, they're fairly close. The one problem is that the script is triggering a NoCanonicalUrl assertion in launchpad, so I can't be certain the measurements are good. If they are, this suggests a 35-40% improvement which would be quite notable.

Just to check, I reran all the different cases, and found all cases involving message data show a similar reduction in run time, whereas cases not involving message data did not see such improvements. Again, the script terminates with an assertion when do_message is enabled, so this performance "improvement" could just be measuring an early termination.

0. Just titles: 5.18 sec ... 4.05 sec
1. do_a: 12.58 ... 15.78
2. do_message: 42.33 ... 37.61
3. do_a + do_message: 51.38 ... 43.31
4. do_fb: 11.68 ... 13.69
5. do_a + do_fb: 17.95 ... 30.55
6. do_a + do_fb + do_message: 1:14.90 ... 59.17
7. do_content: 11.84 ... 11.55
8. do_a + do_content: 24.87 ... 25.30
9. do_a + do_fb + do_content: 25.59 ... 26.82
10. do_a + do_fb + do_message + do_content: 1:08.76 ... 58.82