Device is all black, logcat shows a lot of errors related to l10n: E/GeckoConsole( 76): [JavaScript Error: "TypeError: tokens is null" {file: "app://system.gaiamobile.org/shared/js/l10n_date.js" line: 32}]

E/GeckoConsole( 76): [JavaScript Error: "NS_ERROR_FILE_NOT_FOUND: " {file: "app://system.gaiamobile.org/shared/js/l10n.js" line: 219}] I/power ( 76): *** set_screen_state 1 I/Gecko ( 76): ############################### browserElementPanning.js loaded I/Gecko ( 76): ######################## BrowserElementChildPreload.js loaded E/GeckoConsole( 76): [JavaScript Error: "TypeError: navigator.mozL10n.ready is not a function" {file: "app://system.gaiamobile.org/js/bootstrap.js" line: 92}]

It seems this is only getting reproduced with OTA packages :(. I'm wondering if it may not be related to some recent build system changes on Gaia.

(In reply to Alexandre LISSY :gerard-majax from comment #2) > It seems this is only getting reproduced with OTA packages :(. > > I'm wondering if it may not be related to some recent build system changes > on Gaia. So we've confirmed this doesn't happen if you directly flash the device without doing an upgrade, right?

(In reply to Jason Smith [:jsmith] from comment #3) > (In reply to Alexandre LISSY :gerard-majax from comment #2) > > It seems this is only getting reproduced with OTA packages :(. > > > > I'm wondering if it may not be related to some recent build system changes > > on Gaia. > > So we've confirmed this doesn't happen if you directly flash the device > without doing an upgrade, right? At least, I could not reproduce after doing an APP=system install-gaia on this device. Please note that the logcat error warns about missing l10n_date.js file, but this one is present in the application.zip. I don't have the issue on another device updated via OTA too, but with a build I did yesterday. I'm currently testing updating this device to a build from now, and testing a local revert of commit b6d1c7fae27ec22ee06e3da7898fe2565fc0a8c0 (bug 934727) as suggested by RyanVM on IRC.

Unrelated to bug 934727.

Jason, I can also reproduce after a doing a make clean a then reinstalling gaia only, via install-gaia.

Ok - let me get someone to confirm this on our side. Natalya - Can you have someone flash a trunk build to see if we can reproduce this?

I'm doing bisect on gaia, I have a good lead for now.

Confirmed. A local revert of 6b7a10e5d026cf031b2f3fc0fe5f671ea02dd4f9 makes the system okay: > $ git revert -m 1 6b7a10e5d026cf031b2f3fc0fe5f671ea02dd4f9 > $ make clean && make MOZILLA_OFFICIAL=1 PRODUCTION=1 NOFTU=1 REMOTE_DEBUGGER=1 GAIA_DEV_PIXELS_PER_PX=1.5 LOCALE_BASEDIR=locales/ LOCALES_FILE=locales/languages_all.json APP=system install-gaia So the regression is likely coming from bug 968661

Alright - let's get that patch backed out then.

Bug 968661 reverted.

Hi Alex, with my commit, I cannot see the log you mention in comment 0 and comment 1 by > $ make clean && make MOZILLA_OFFICIAL=1 PRODUCTION=1 NOFTU=1 REMOTE_DEBUGGER=1 GAIA_DEV_PIXELS_PER_PX=1.5 LOCALE_BASEDIR=locales/ LOCALES_FILE=locales/languages_all.json APP=system install-gaia and also, system app works fine. Maybe I can also have your build version?

(In reply to George Duan [:gduan] [:喬智] from comment #12) > Hi Alex, > with my commit, I cannot see the log you mention in comment 0 and comment 1 > by > > $ make clean && make MOZILLA_OFFICIAL=1 PRODUCTION=1 NOFTU=1 REMOTE_DEBUGGER=1 GAIA_DEV_PIXELS_PER_PX=1.5 LOCALE_BASEDIR=locales/ LOCALES_FILE=locales/languages_all.json APP=system install-gaia > > and also, system app works fine. > Maybe I can also have your build version? After investigating a lot with George on IRC, we found out that the error was triggered by the change of l10n.js: my gaia clone was quite an old one, from the time where we had a apps/NAME/shared/ directory populated. This made the build system changes getting confused and pulling this into build_stage/, hence breaking the whole.