GNOME Bugzilla – Bug 556327
A population with rev #3238 is much slower than libgda rev #3208
Last modified: 2008-10-20 06:50:45 UTC
Please describe the problem: Anjuta's sqlite-backend population is 10-15 times slower with rev 3238 compared to rev 3208. The apis didn't change in Anjuta's module, so it's probably a libgda problem. Steps to reproduce: 1. Test svn-head Anjuta against the two mentioned libgda revisions. Actual results: Expected results: Does this happen every time? Other information:
Can you check with rev #3241?
(In reply to comment #1) > Can you check with rev #3241? > still present. I had a check with rev #3225 and the bug was already present.
(In reply to comment #2) > (In reply to comment #1) > > Can you check with rev #3241? > > > > still present. > I had a check with rev #3225 and the bug was already present. > ok, seems like the binary search went good :) rev #3220 is bug-free, and the speed is ok. Looking at the changelog seems like it's rev #3221 which caused the problem.
Ok, thanks for the hint, I'll investigate this ASAP.
I've got some more infos: here it is the debug-printouts of the timing with the latest revision and with 3220. ------------- 3242 ** Message: elapsed: 0.447980 for (13) [0.034460 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 1.038639 for (16) [0.064915 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 19.700733 for (183) [0.107654 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 3.496080 for (15) [0.233072 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 9.078364 for (33) [0.275102 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.677795 for (2) [0.338897 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 13.963586 for (36) [0.387877 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 4.118010 for (12) [0.343167 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 6.390777 for (17) [0.375928 per symbol] ** Message: sdb_engine_populate_db_by_tags () ------------- 3220 ** Message: elapsed: 0.429302 for (13) [0.033023 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.413188 for (16) [0.025824 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 2.774804 for (183) [0.015163 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.401142 for (15) [0.026743 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.534034 for (33) [0.016183 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.189534 for (7) [0.027076 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.836386 for (31) [0.026980 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.245796 for (12) [0.020483 per symbol] ** Message: sdb_engine_populate_db_by_tags () ** Message: elapsed: 0.422233 for (17) [0.024837 per symbol] the files are the same.
I'm currently working on resolving this bug. I'm using the benchmark program in the plugins/symbol-db/test dir. Thanks for helping identifying the revision where the problem came from.
I've observed some drastic performances improvements with rev #3244, can you check?
yeah it's flying again! I think this bug is fixed. Thanks for you kind support, Massimo
Great! So I'm closing this bug.