789529 - Updater tests can't find channel-prefs.js

Status: RESOLVED FIXED

(Firefox :: General, defect)

Description

[Jim Mathies [:jimm]]

channel-prefs.js ends up in:

dist/bin/browser/defaults/pref/channel-prefs.js

which isn't found by update tests.

Moving it to 'preferences' solves the problem. But based on the comments in bug 756325 maybe that's not the right fix.

cc'ing a few people.

Comment 1

[Jim Mathies [:jimm]]

The code in the app makefile was updated to:

ifndef LIBXUL_SDK
# channel-prefs.js is handled separate from other prefs due to bug 756325
libs:: $(srcdir)/profile/channel-prefs.js
	$(NSINSTALL) -D $(FINAL_TARGET)/defaults/pref
	$(PYTHON) $(topsrcdir)/config/Preprocessor.py $(PREF_PPFLAGS) $(ACDEFINES) $^ > $(FINAL_TARGET)/defaults/pref/channel-prefs.js
endif


FINAL_TARGET points to the browser sub dir.

Comment 2

[Jim Mathies [:jimm]]

Moving this file back down to dist/bin also solves the problem. It also lets both browser share the same channel-pref file, so maybe that's the right fix?

Comment 3

[Robert Strong (they/them - no direct email)]

This is filed as Win7 but isn't this a Win8 bug?

The channel-prefs.js needs to end up in the defaults/pref directory that is read by the app per bug 756325.

Nick, I'm still tempted to get rid of channel-prefs.js entirely and go with allowing QA to provide an override other than the channel-prefs.js file. If you ever need to change the channel that would still be possible when updating.

Comment 4

[Robert Strong (they/them - no direct email)]

(In reply to Jim Mathies [:jimm] from comment #2)
> Moving this file back down to dist/bin also solves the problem. It also lets
> both browser share the same channel-pref file, so maybe that's the right fix?
I think that is the better way to go for now. If nothing else, they are the same install and there shouldn't be the possibility of having two different channels based due to two channel-prefs.js files.

Comment 5

[Jim Mathies [:jimm]]

(In reply to Robert Strong [:rstrong] (do not email) from comment #3)
> This is filed as Win7 but isn't this a Win8 bug?

This is fall out from the metro-build bug, so once that land on mc this will show up there.

> I think that is the better way to go for now. If nothing else, they are the
> same install and there shouldn't be the possibility of having two different
> channels based due to two channel-prefs.js files.

Ok, I'll land that change on elm to be sure it addresses the update test failures. Will post a patch here as well for Glandium.

Comment 6

[Jim Mathies [:jimm]]

Attachment: patch

This just reverts a single change in the "Move browser application in a subdirectory (PoC)" patch in bug 755724.

Comment 7

[Jim Mathies [:jimm]]

Attachment: packager

revert packager changes as well.

Comment 8

[Jim Mathies [:jimm]]

Attachment: packager

Comment 9

[Mike Hommey [:glandium]]

Comment on attachment 659304 [details] [diff] [review]
patch

This is needed on m-c, as bug 787180 landed there a while ago. This is effectively a noop until FINAL_TARGET != DIST/bin.

Comment 10

[Mike Hommey [:glandium]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/4fe60dd3d145

Comment 11

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/4fe60dd3d145