User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:18.104.22.168) Gecko/20101203 Firefox/3.6.13 Build Identifier: Bugzilla 4.0 / tip using "confirm match" leads to match failed error due to an added comma. " Assignee: firstname.lastname@example.org, did not match anything " tested on landfill 4.0 and tip . Reproducible: Always Steps to Reproduce: 1. add a non-unique string in the assignee field 2. don't use the autocomplete 3. in the "confirm match" screen, choose an assignee Actual Results: Assignee: email@example.com, did not match anything Expected Results: Assignee choosen in the drop-downbox is accepted
Let's mark it as a blocker for now. I will investigate when I'm back home in an hour or so. mkanat, let me investigate before we release 4.0 as I suspect it's a regression.
Flags: blocking4.0? → blocking4.0+
Target Milestone: --- → Bugzilla 4.0
Version: unspecified → 4.0
Status: UNCONFIRMED → NEW
Ever confirmed: true
I haven't researched to figure out where this was introduced, but there seem to be two problems: 1) User::match is looking for multiple values for fields where there can only be one value. 2) confirm-user-match was sending back an empty "assigned_to" (etc.) as part of hidden-fields.html.tmpl, which meant that there were two assigned_to fields on the page. I solved #2, because that seemed like a simple fix, and it seemed like a clear bug that confirm-user-match was sending back an extra, empty field value.
Assignee: create-and-change → mkanat
Status: NEW → ASSIGNED
Attachment #512478 - Flags: review?(LpSolit)
This was most likely regressed by bug 585802, although in a sense, it only exposed a bug in confirm-user-match.
The bug in confirm-user-match was hidden before because we joined fields by a space--putting an extra space at the end of the assigned_to or qa_contact fields didn't matter. But now we only accept commas as separators, and adding a comma does matter.
(In reply to comment #5) > 1) User::match is looking for multiple values for fields where there can only > be one value. We force it to take all values into account even with type='single', because we don't want to choose the (wrong) value randomly. I spent enough time on this method in Bugzilla 2.20 and in bug 534057 to know that's the correct behavior. :) > 2) confirm-user-match was sending back an empty "assigned_to" (etc.) as part of > hidden-fields.html.tmpl, which meant that there were two assigned_to fields on > the page. This is indeed the reason of the problem, yes. Nice catch! :)
Comment on attachment 512478 [details] [diff] [review] v1 Works fine. Thanks. r=LpSolit
Attachment #512478 - Flags: review?(LpSolit) → review+
Committing to: bzr+ssh://bzr.mozilla.org/bugzilla/trunk/ modified template/en/default/global/confirm-user-match.html.tmpl Committed revision 7717. Committing to: bzr+ssh://bzr.mozilla.org/bugzilla/4.0/ modified template/en/default/global/confirm-user-match.html.tmpl Committed revision 7557.
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.