Thunderbird Meeting Details
- nth10sd & wsmWK: Bug day on Thursday 24 April 2008.
- nth10sd will be present for
all threetwo timeslots, sessions 2 and 3.
- wsmWK will be present for all three timeslots.
- nth10sd will be present for
- Prepare Thunderbird 3 Alpha 1 release notes for the upcoming release.
- nth10sd has the pre-first draft out.
- Please comment on how we can improve the release notes.
- Code freeze tonight at 2359h PDT -- Code freeze is postponed until 2359h PDT Wed 23rd April
- For Thunderbird 3 Alpha 1.
- Everyone: a code name for non-release TB builds.
- Previous bug day (17 April 2008)
- Low turnout.
- But high numbers of triaging per volunteer / developer.
- We are still in triple digit numbers of bugs changed.
- Let's keep up the momentum!
- Rick (on last week's TODO item #1): FF schedule affects TB schedule but 188.8.131.52 is available for testing, message has been sent out. There were some issues, got them resolved in time to QA to start. His focus on making 2.x out the door (approx 2 days), then he will work with Thomas to figure out what is needed for 3.0a1, especially for automated builds - buildbot (no estimation for this work yet). Hardware is ready for TB builds.
- Last week's TODO item #2: FF uses the same name (minefield) for non-release builds to emphasize the problems the user might have. We should use the same idea for TB.
- Calendar meeting: Meeting results can be found at meeting page.
- Google SoC page: Mozilla projects
- Driving (Highlights from Dmose's recent message sent to newsgroups)
- Since two of the drivers are going to be in transit for the next day or so, and since there is still a relatively small amount of bug work left to do, we decided at today's Thunderbird weekly meeting to slip the code freeze by one day to Wednesday, April 23, 23:59 Pacific.
- The intent is that at that point, we'll put verbiage on the Tinderboxen forbidding checkins to mail/ and mailnews/ of non-3.0a1-blocking bugs without explicit approvals. Those directories will remain closed until we've cleared the blocker list (either by fixing or cutting remaining bugs), spun builds, and tagged the release.
- Our goal is not to keep the tree stay closed too long.
- Our intent is to build a release checklist as we go, so that this stuff is written down somewhere, possibly either updating or starting from http://developer.mozilla.org/en/docs/Mozilla_Release_Checklist. http://wiki.mozilla.org/Firefox3/Beta2ReleaseChecklist looks like useful source material as well.
- Thunderbird 184.108.40.206 build is underway ... I managed to mangle the Windows build tho. nthomas did a fair bit of cleanup this AM. It is now at the signing stage of the build.
- resourcing (qa, machines) for Thunderbird 3.0a1 will be trumped by any FF build needs
asuth (Andrew Sutherland):
- blockers worked on:
- opaque drag-and-drop gtk2 thunderbird blocker bug 422700 fix by using the workaround provided by bug 376238. Implemented a haunted prototype fix for bug 376238. Haunted in that the transparent dragging would sometimes work the first time, but never any time after that. Spent a fair amount of time trying to figure out why this happened, but eventually gave up.
- command-shift-M for move again busted on OS X blocker bug 428062 fix by bumping it to option-command-M; there was a collision with command-shift-M for new message. Much thanks to philor.
- random bugs (bug-day) that got a lot of focus:
- investigated why text/html paste from netbeans didn't work on bug 395218, turns out the gtk2 widget nsClipboard code really wants to see UTF-16 0xfe 0xff hint bytes or an explicit charset in the payload or it ignores it. Also, it doesn't consider the possibility of targets with explicit charsets in the actual target atom. Moved to core/widget:gtk.
- performance work:
- currently generating a trace against Standard8's bloat-test automation, but (of course) with all the bloat test stuff turned off.
- figured out why dtrace probes were double-firing bug 403132; turns out the Apple linker magic gets confused with -dead_strip enabled. Only documented the work-around thus far; will at least provide an ugly patch.
- in the process of trying to ensure accuracy of time-attribution by using/augmenting the patch from bug 407542 so we get XPConnect js_Invoke calls too. General accuracy-wise, it looks like the function names reported (when they are reported) are sufficiently unreliable that it's better to just use the line-numbers that are (more) reliably reported and we can use thanks to our quasi-fancy parsing. (Based on existing bugs like bug 422864, it sounds like a non-trivial fix for me to pursue for now.)
- The system architecture documentation effort continues. More diagrams and description will be added this week. You can check the diagrams at System Diagrams
- Started to work with Dale R. Wiggins (Penelope project) to support selective message segment downloads, and to improve offline mode - imap.
- Work proposal: Developing a protocol test harness to test:
- Message Storage
- Calendar F2F meeting.
- Performance test improvements
- GetSubFoldersObsolete removal (bug 420614 ongoing).
- Calendar F2F meeting
- Various reviews
- 3.0a1 driving
davida (David), dmose (Dan), Standard8 (Mark) , rick, asuth (Andrew), clarkbw (Bryan), beckley (Jeff), emre, ??