QA/Firefox3/TestResults/RC1/RC2BugCandidates

« QA/Firefox3/TestResults/RC1

Bug Candidates for Triage

If the following bugs are strong candidates for building a RC2, please set the following in bugzilla:

• set whiteboard to "[RC2?]"
• nominate the blocking-firefox3 or blocking1.9 flag
• or... nominate the wanted 1.9.0.x flag

Bugs (as of 5/23/08):

• bug 433340 – bookmark dialog covers candidate window when using IME. Status: recommended for RC2 with patch in place.
• Filed bug 433440 – Setting Fx to default browser on installer does not launch browser. Looks like a dupe of bug 395891 and bug 295248
• Renom'd bug 433298 EULA cut off checkbox/buttons on bottom - unusable on small (EeePC) screens for blocking-firefox3. Linux users in this mode have a minimal way to move forward with the installer. Duped bug 433627 in the process.
• bug 433796 JavaScript component does not have a method named: "observe"' when calling method: [nsIObserver::observe]
• bug 433748 Error: this._processor is null
• bug 433370 – NS_ERROR_ILLEGAL_VALUE in [nsITreeView::getCellText]
• I get a page alert and a dialog coming down on SSL Error exceptions. Cannot find the bug but the issue is
• Favicons are not imported from Safari bug 326701
• bug 420449 (unable to drag folders in bookmarks toolbar)
• bug 433795 (error in the console when trying to move folder)
• bug 421775 (moving items in the bookmarks library)
• bug 433934 (Name field of History Entries should not look like it can be edited)
• bug 433959 (Page info for mozilla.com shows a cnn.com cookie)
• bug 433975 (session restores to wrong session)
• bug 433942 – Focus lost when going "Back" a page after dismissing a notification bar
• bug 433605 - NS_ERROR_NOT_AVAILABLE [nsIChannel.contentType]
• Confirming during FFT: bug 324334 - Cannot switch to Error Console in the "Window" menu
• bug 419928 – Add "InstallerLocation" for Java in PFS. PFS is broken on production, stuck in infinite loop. Webdev fix by adding a manual install option.

Crashers

• bug 433525 - crash [@ nsNavHistoryQueryResultNode::IsContainersQuery()]. Number 1 topcrash, is a recent regression which we should probably respin for. The bug has a patch ready to go.
• bug 434401 – crash [@ gfxWindowsFont::GetOrMakeFont(FontEntry*, gfxFontStyle const*)]. that may be happening in thebes or layout. dbaron believes thebes.
• bug 434403 – startup crash [@ nsDocShell::SetupNewViewer(nsIContentViewer*)]. recent regression. While we saw this crash in beta 5 (presumably from upgrading from beta 4), it's much greater this release and happens at a different address.

Accessibility

• Filed bug 434491 – Several ns*Accessible classes leak on simple paragraph with 1 link in it. Status: Assigned to jonas
• bug 432970 - Shutdown() of nsXULTooltipAccessible is not called. Affects Linux only, but can cause accessible tree corruption because tooltip accessibles from tabs that are already closed may still be around and accidentally accessed by the Orca screen reader. Again not a blocker, but a good candidate to tag along.
• bug 434002 - event show isn't fired for treecol accessible. This is an embarrassing oversight from review of fix for bug 413777. It basically means that event_show is never ever called for any accessible even if it needs to be. The most obvious use case we found isn't in Firefox but Thunderbird, when column
• bug 432467 - firefox segfaults in plone kupu editor [@ nsDocAccessible::FlushPendingEvents]. This is a crasher reported by the RedHat folks that indicates a race condition between two things trying to release the same accessible. See comment #4 for the description. Assessment of risk is also in the bug. I wouldn't respin for this, but if it could tag along if RC2 is being done anyway, that would be very good I think. This can happen with other ajax-enabled stuff as well.
• bug 432869 - Divide by zero in IsProbablyForLayout(). This came up while triaging crash-stats. It is a simple check for certain types of layout tables to make sure we never divide by zero. Was probably missed during review of another bug's patch. Again, for safety reasons, it would be good to have this, but not blocking the release if 3.0.1 is only 4 to 6 weeks away after that. headers of the message trees change when switching from folder to folder. But this can happen to any extension suing tree tables in Firefox and replacing one view with another and changing column headers, and it can happen to some types of Ajax-empowered web sites, leading to accessible tree corruptions. Again not a blocker per se, but again a good candidate to ride along on an RC2 build.