| Title | Jobspec not suitable for creating new jobs |
| Status | closed |
| Priority | essential |
| Assigned user | Gareth Rees |
| Organization | Ravenbrook |
| Description | The P4DTI creates a Perforce jobspec that's fine for jobs which are replicated from the defect tracker, but not so good for jobs that are created in Perforce. |
| Analysis | [1] offers the following suggestions for the Bugzilla integration: P4DTI: Reporter word 255 optional , no default Suggested: Reporter word 255 required, default $user P4DTI: Creation_Timestamp date 20 optional Suggested: Creation_Timestamp date 20 once P4DTI: Update_Timestamp date 20 optional Suggested: Update_Timestamp date 20 always The reporter suggestion looks good, but it needs to be "default", not "required" otherwise we get into trouble when replicating a Bugzilla bug with no reporter. We can't have any "once" fields, because they break the refresh script [2] asks that in the TeamTrack integration we handle Submit_Date, Last_Modified_Date, and Submitter. [NB 2002-04-05] This is actually a complex set of interrelated problems. Firstly we have the problem that the Bugzilla integration will fail to migrate jobs and/or to replicate new Perforce jobs if certain fields are in the replicated_fields. For Bugzilla this is job000493 (delta_ts) job000494 (qa_contact), and job000497 (lastdiffed). Secondly we have the problem that neither the Bugzilla nor the TeamTrack integrations will replicate new Perforce jobs with the defaults values of certain fields. For Bugzilla this is job000491 (reporter, qa_contact). For TeamTrack, this is job000501 (LASTMODIFIEDDATE, SUBMITDATE). Thirdly, we have the problem that the automated test suite doesn't handle the replication of some fields very well. This is job000492 (reporter), and job000495 (qa_contact). Fourthly, we have the problem that migration breaks for Bugzilla if some "numeric" fields are replicated. This is job000496. Lastly, we have the problem that some "numeric" fields from Bugzilla don't translate to Perforce very well. This is job000498. |
| How found | customer |
| Evidence | [1] <http://info.ravenbrook.com/mail/2001/12/14/17-17-48/0.txt>[2] < http://info.ravenbrook.com/mail/2002/01/08/01-12-56/0.txt> |
| Observed in | 1.3.3 |
| Introduced in | 1.3.0 |
| Created by | Gareth Rees |
| Created on | 2001-12-14 18:03:27 |
| Last modified by | Nick Barnes |
| Last modified on | 2002-04-05 16:47:25 |
| History | 2001-12-14 GDR Created. 2002-01-08 GDR Extended scope of defect to cover TeamTrack; added [2]. 2002-04-05 NB Analysed, breaking down into a number of other jobs. |
| Change | Effect | Date | User | Description |
|---|---|---|---|---|
| 27878 | closed | 2002-04-05 16:35:27 | Nick Barnes | Handle replication of new jobs when LASTMODIFIEDDATE or SUBMITDATE are in replicated_fields. |
| 27829 | open | 2002-04-03 17:44:03 | Nick Barnes | migrate to Bugzilla even when replicating numeric fields. |
| 27822 | open | 2002-04-03 14:03:13 | Nick Barnes | job000494: allow user of 0 for qa_contact, but prohibit it elsewhere. |
| 27821 | open | 2002-04-03 13:55:32 | Nick Barnes | job000493: fix timestamp translator of empty strings. |
| 27800 | open | 2002-04-02 16:44:44 | Nick Barnes | User fields in the jobspec should default to $user. |