538511 - release automation updates for maemo5

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P3)

Description

[Aki Sasaki (not active)]

Hoping this is close to Maemo4, but it won't be free.

Comment 1

[Aki Sasaki (not active)]

Tested my patches on sm02; I'll clean those up for review after jhford lands his massive patch tomorrow.

One thing I noticed, in the partner repack (which is busted for maemo5-gtk atm because of this) is that maemo5-gtk puts its bits in /usr/local/lib instead of /opt/mozilla.

I think this is a Fennec bug, but I need to confirm.

Comment 2

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to comment #1)
> Tested my patches on sm02; I'll clean those up for review after jhford lands
> his massive patch tomorrow.
> 
> One thing I noticed, in the partner repack (which is busted for maemo5-gtk atm
> because of this) is that maemo5-gtk puts its bits in /usr/local/lib instead of
> /opt/mozilla.
> 
> I think this is a Fennec bug, but I need to confirm.

Looks like the mozconfigs are missing the install location lines, so we are defaulting:
http://hg.mozilla.org/build/buildbot-configs/file/a05483b93162/mozilla2/mobile/maemo4/mobile-browser/nightly/mozconfig#l7

You need both of those lines

Comment 3

[Aki Sasaki (not active)]

(In reply to comment #2)
> Looks like the mozconfigs are missing the install location lines, so we are
> defaulting:
> http://hg.mozilla.org/build/buildbot-configs/file/a05483b93162/mozilla2/mobile/maemo4/mobile-browser/nightly/mozconfig#l7
> 
> You need both of those lines

Aha, nice catch. Thanks, I'll add that to my patch.

Comment 4

[Aki Sasaki (not active)]

Attachment: maemo5 release repacks -- more qt tar.gz

Comment 5

[Aki Sasaki (not active)]

Attachment: maemo5-gtk, maemo5-qt release builder configs

1) add prefix+libdir to all maemo5 mozconfigs
2) add maemo5-qt/mobile-browser/release/mozconfig that is the same as the nightly, but with official branding
3) partner repack builders per platform at this stage, since trying to do maemo5-gtk and maemo4 in the same pass had some issues (same filename)

Looks like I did most of the heavy lifting pre-1.1rc1; it's largely a matter of just turning these on :)

Comment 6

[Aki Sasaki (not active)]

Attachment: testing stage release_mobile_config

For reference+posterity. This is the release-fennec-2.0.py that I used for my testing.  It's got Android in it, so requires the patches that I'll be posting to bug 561901 (or just remove android-r7 from the platform list).

Since Android and Maemo5-qt are non-1.9.2 platforms, I built off of trunk.
QT was not a partner repack (and most likely we'll only deliver one deb to partners, so the partnerRepackPlatforms will stay at 1 platform for the time being). android-r7 was left out of the l10nPlatforms til we have localized Android builds.

Comment 7

[John Ford [:jhford] CET/CEST Berlin Time]

Comment on attachment 453802 [details] [diff] [review]
maemo5-gtk, maemo5-qt release builder configs

Looks good.

My only concern is the 4 (?) places where this line occurs.

>+            sb_target=pf.get('sb_target', 'CHINOOK-ARMEL-2007'),

I don't think we should be using a default or {}.get in this case.  I would much rather find out that i didn't set a scratchbox target on checkconfig than find out we shipped a FREMANTLE_ARMEL browser to chinook users.  If we really do need to have a default for some reason, I think that FREMANTLE_ARMEL is a more appropriate default.

r+ if the .get( is replaced with pf[.

Comment 8

[Aki Sasaki (not active)]

Comment on attachment 453802 [details] [diff] [review]
maemo5-gtk, maemo5-qt release builder configs

checked in, with pf['sb_target'].

http://hg.mozilla.org/build/buildbot-configs/rev/3ba766d66489

Comment 9

[Aki Sasaki (not active)]

Comment on attachment 453798 [details] [diff] [review]
maemo5 release repacks -- more qt tar.gz

http://hg.mozilla.org/build/buildbotcustom/rev/642e73401a4a