| Title | Bugzilla integration fails to update timestamp if only description changes |
| Status | suspended |
| Priority | optional |
| Assigned user | Nick Barnes |
| Organization | Ravenbrook |
| Description | In Bugzilla, comments and descriptions are kept in the 'longdescs' table, separate from the 'bugs' table. When the P4DTI replicator updates this table for a Bugzilla bug (because a Perforce user has changed the 'Description' field of the corresponding Perforce job), it fails to update the delta_ts timestamp of the bug itself. |
| Analysis | In bugzilla.py, the update_bug method skips the call to update_row if there are no updates. It needs to explicitly say "update bugs set delta_ts = now()" (or similar). [NB 2002-03-28] But if it does this then it can't distinguish longdesc updates made by the P4DTI from those made by other users (because longdesc updates don't appear in the bugs_activity table), which means that we get conflicts if either (a) using a pre-2002.1 P4D or (b) running successive polls by hand with poll.py. So we need to keep this behaviour for now. Downgraded to 'optional'. [NB 2002-03-28] Entries in bugs_activity are only used in Bugzilla for showing bug activity tables, and it would seem to be harmless for the P4DTI to update bugs.delta_ts and bugs_activity (and p4dti_bugs_activity), which would fix this bug without introducing conflicts. Checking this by asking the mozilla-webtools list. |
| How found | inspection |
| Evidence | I just know. |
| Observed in | 1.4.0 |
| Introduced in | 0.5.0 |
| Created by | Nick Barnes |
| Created on | 2002-03-25 17:20:45 |
| Last modified by | Nick Barnes |
| Last modified on | 2018-07-05 17:27:57 |
| History | 2002-03-25 NB Created. 2002-03-28 NB Added several notes on potential problems. 2018-07-05 NB Suspended because the P4DTI is obsolete. |