Bugzilla – Full Text Bug Listing
| Summary: | PostgresCatalogDatabase.refresh() implemented incorrectly | ||
|---|---|---|---|
| Product: | [Tools] Data Tools | Reporter: | Brian Vosburgh <brian.vosburgh> |
| Component: | Enablement | Assignee: | Pierre Queinnec <pierre.queinnec> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | pierre.queinnec |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | All | ||
| Whiteboard: | |||
Hi Brian, You're right. I'll address this shortly. By the way, I notice the MySQL enablement does the same thing. Do you also see the problem with MySQL? Committed and pushed for Postgres. See http://git.eclipse.org/c/datatools/org.eclipse.datatools.enablement.postgresql.git/commit/?id=4ca98c58b0664950d12a35ed9558c75453a78fff I didn't touch the MySQL part; please tell me from the Dali point of view if we should also correct it. There plenty of problems with MySQL. :-) See the following: bug 279721 bug 279726 bug 201957 bug 249013 |
The method PostgresCatalogDatabase.refresh() is implemented incorrectly. It must call super.refresh() and, as a result, need *not* call RefreshManager.referesh(...). Since super.refresh() is not called, the database's catalogs are never reloaded, leading to some failed Dali tests. Unfortunately, WTP is not on Git yet; so I'm not set up to give you a patch; but the modified method should look like this: public void refresh() { synchronized (AUTHORIZATION_IDS_LOCK) { if (authorizationIdsLoaded.booleanValue()) { authorizationIdsLoaded = Boolean.FALSE; } } super.refresh(); } It appears this regression (i.e. refresh() *used* to work with a PostgreSQL database) occurred with the changes associated with bug 359861.