Releases/Firefox 24.0b1/BuildNotes
who: mgerva bugs hit: * 902112 Tracking bug for build and release of Firefox & Fennec 24.0b1 * 901892 release sanity broken by mozconfig changes * 902066 fx 24.0b1 release fails running make -f client.mk build * 902084 firefox 24.0b1 l10n repacks for macosx64 fail * 902091 failed l10n repacks for firefox 24.0b1 on windows
build 1
Failed steps
- release verification fails because a different options in mozconfig files, android/nightly and andorid/releases. Filed bug 901892 and proposed a patch.
- {firefox|xulrunner|fennec}_source failed bug 902066
configure: error: Library requirements (gtk+-2.0 >= 2.10.0 gtk+-unix-print-2.0 glib-2.0 gobject-2.0 gdk-x11-2.0) not met; consider adjusting the PKG_CONFIG_PATH environment variable if your libraries are in a nonstandard prefix so pkg-config can find them. configure:16945: checking for gtk+-2.0 >= 2.10.0 gtk+-unix-print-2.0 glib-2.0 gobject-2.0 gdk-x11-2.0 Fix above errors and then restart with "make -f client.mk build"
- linux repacks bug 902066
configure:16945: checking for gtk+-2.0 >= 2.10.0 gtk+-unix-print-2.0 glib-2.0 gobject-2.0 gdk-x11-2.0
- win32 repacks 902091
client.mk:90: *** Pymake is required to build on Windows. Run |./mach build| to automatically use pymake. Or, invoke pymake directly via |python build/pymake/make.py|.. Stop.
- macosx64 repacks 902084
fatal error: too many errors emitted, stopping now [-ferror-limit=] 20 errors generated. subprocess.CalledProcessError: Command '['make', 'tier_base']' returned non-zero exit status 2
build 2
Bug 851606 - Disable bug 818340 for Beta until we're ready
mail from Bhavana: missed landing in Firefox 24 Beta prior to go-to-build for beta 1 & this is an important enough issue to cause a re-spin for Desktop and Mobile.
Kicked off without source/windows repacks/osx repacks fixes. We'll have to rebuild those builders when the fixes land + are tagged.
- bug 902066 fixed linux32 l10n repack, and the source jobs - all finished.
- bug 902084 fixed mac l10n repacks, all finished
- https://releng.etherpad.mozilla.org/180
Timeline * 2013-08-05 22:01 PDT RelMan sends initial 24.0b1 build 1 go-to-build * 2013-08-05 21:55 PDT Release runner failed * https://bugzilla.mozilla.org/show_bug.cgi?id=901892 * 2013-08-06 05:37 PDT fixed * 2013-08-06 11:49 PDT resolved * 2013-08-06 05:43 PDT Tagging started * 2013-08-06 06:42 PDT Source fails * 2013-08-06 08:30 PDT L10n repacks start failing (osx) * 2013-08-06 09:56 PDT L10n repacks start failing (linux) * 2013-08-06 10:38 PDT L10n repacks start failing (win32) * 2013-08-06 11:26 PDT Bug filed for source, linux repacks * https://bugzilla.mozilla.org/show_bug.cgi?id=902066 * 2013-08-06 14:42:18 PDT mozconfig fix landed (compiler/env change) * 2013-08-06 20:06 PDT signing servers updated with increased gpg size * 2013-08-06 21:49 PDT firefox source completed successfully (24.0b1 build2) * 2013-08-06 22:43 PDT xulrunner source completed successfully (24.0b1 build2) * 2013-08-06 22:43 PDT fennec source completed successfully (24.0b1 build1) * 2013-08-06 11:45 PDT Bug filed for macosx repacks * https://bugzilla.mozilla.org/show_bug.cgi?id=902084 * 2013-08-07 02:25 PDT Patches landed, merged, tagged, rolled out with buildbot reconfig (clang install via tooltool, compiler/env change for mozconfigs) * 2013-08-07 03:38 PDT macosx repacks start going green * 2013-08-07 03:57 PDT macosx repacks finish going green * 2013-08-06 11:55 PDT Bug filed for win32 repacks * https://bugzilla.mozilla.org/show_bug.cgi?id=902091 * 2013-08-06 15:09 PDT Fix written+reviewed for using pymake; r+ed contingent on green staging run * 2013-08-06 17:55 PDT Corresponding buildbot{-configs, custom} patches written + posted to bug. * Many hours of busted staging builds + debugging + attempted fixes * 2013-08-07 05:04 PDT Staging still busted ("State of 24.0b1" email) * After switching to pymake, we hit the tier_base compilation error. Removing tier_base from the makeDirs list moved the error down into tier_nspr; didn't fix. * This error seemed to show we were losing the msvs10 env specified in the mozconfig * This became https://bugzilla.mozilla.org/show_bug.cgi?id=903118 but we found a workaround. * We dropped this as a potential fix at this point to try our fix in bug 898099 (switching to w64 builders) * 2013-08-06 16:51 PDT "Possible FF24.0b1 delay" email to r-d * 2013-08-07 06:17 PDT Switch to w64 builders (first bug activity since 2013-07-26, but had been working on it earlier) * https://bugzilla.mozilla.org/show_bug.cgi?id=898099 * Attempt to make release repacks closer to nightly repack setup * 2013-08-07 15:50 PDT upload_dir makes it to production * 2013-08-07 21:41 PDT patches reviewed, ready to land, but want to avoid busting 23.0.1 chemspill. Testing towards that proceeds overnight * 2013-08-08 11:38 PDT We switch to w64 builders * This unfortunately fails to fix the problem. Uncaught in staging because we were testing against 23.0b8 * 2013-08-08 13:09 PDT pymake and windows l10n release repacks don't get along * https://bugzilla.mozilla.org/show_bug.cgi?id=903116 * Revisiting the progress made in bug 902091 * 2013-08-08 17:25 PDT Four backslashes doesn't work * 2013-08-08 17:48 PDT Initial EIGHT BACKSLASHES patch gets further * 2013-08-08 18:11 PDT Verified switching to gmake mid-run doesn't work * 2013-08-08 19:08 PDT Written + tested + reviewed EIGHT BACKSLASHES patch * 2013-08-08 23:10 PDT Unable to get past mar issues after many staging runs, broken updates * Decision made to proceed without windows partial updates; work had been going on towards that in parallel with EIGHT BACKSLASHES patch * 2013-08-09 09:27 PDT We proceed without windows partial updates * 2013-08-09 throughout the day: we attempt to replicate the mar bustage in staging without success. * 2013-08-09 03:33 PDT Provide way to ship 24.0b1 updates without having win32 l10n * https://bugzilla.mozilla.org/show_bug.cgi?id=903357 * Filed after work had already proceeded on this * 2013-08-09 05:32 PDT "Specifics for serving 24.0b1 installers if win32 l10n doesn't happen" email * 2013-08-09 09:54 PDT "Firefox 24.0b1 build2 is avalaible on betatest" email * 2013-08-09 18:08 PDT "Firefox 24.0 Beta 1 SHIPPED" email * 2013-08-09 22:29 PDT Firefox 24 beta 1 xml parse error on startup: browser.xul, devtoolsMenuBroadcaster_ChromeDebugger * https://bugzilla.mozilla.org/show_bug.cgi?id=903716 * Also fallout from pymake * Theory that the mozconfig l10n dir path is wrong -- found this wasn't true * 2013-08-12 16:09 PDT LOCALE_MERGEDIR fix via EIGHT BACKSLASHES * verified fixed in staging * 2013-08-12 16:55 PDT 24.0b2 build 1 go-to-build, without 24.0b1 partials
build 3 (mobile only)
mail from Bhavana:
[Mobile Only] Upcoming : Request for Firefox 24 Beta 1 build #3;
The backout for the feature is being tracked in Bug 902783 - Fx24 Beta: back out Bug 866957 and Bug 877725. :mfinkle is currently working on the patch but will be able to land it on the Beta tree by tomorrow morning once the code is reviewed. So I am expecting to send the go-to-build request around morning PT as soon as there is a clean landing.
- per email, build 3 posted to google play and pushed to mirrors