On Jan 27, 2011, at 9:34 AM, Marius Dumitru Florea wrote:
On 01/26/2011 04:41 PM, Vincent Massol wrote:
Hi devs,
Sorin and Silvia are helping us in term of improving the quality of platform/XE/XEM by systematically testing milestones/RCs/final versions.
They've started a TestReports space on xwiki.org where he's going to put all his test reports from now on: http://www.xwiki.org/xwiki/bin/view/TestReports/
In those reports, when a test fails they'll need to create a JIRA issue (instead of pointing to a page on the incubator). This would make it much easier for reporting purpose, to see all the issues they've created and for us to see all issues found by the Q&A team.
The problem is that our current JIRA strategy is to NOT create jira issue for issues that haven't been "released" already (instead the strategy is to comment in them).
I'm proposing to relax this rule in the following manner:
* Allow to have "affects" and "fix for" be the same value in JIRA * Automatically exclude from our release notes issues where "affects" == "fixfor"
In addition IMO it would be good for Sorin/Silvia to:
* Link the newly created jira to an existing jira
WDYT?
+1 provided it is used as a last resort, after:
* making sure the issue is not present in the latest release
if we find the issue already exists we'll close it with a duplicate.
* making sure there aren't any automated tests failing for the same reason (-> devs are already notified)
Agreed, that said if functional tests are failing they need to be fixed and fixing them will require a commit usually, which can be done against the created issue. But I agree we should try to minimize those issues as they're not necessary.
* asking on IRC or mailing list if there were recent changes to the module affected by the issue (-> comment on existing issue)
Thanks -Vincent