1224901 - Improve travis for b2g manifest

Status: RESOLVED FIXED

(Firefox OS Graveyard :: GonkIntegration, defect)

Description

[:gerard-majax]

+++ This bug was initially created as a clone of Bug #1224861 +++

We need to improve manifest checking within travis. I suggest we should at least add trying to read all manifests. That would have triggered the bell for avoiding bug 1224861.

I don't know if we can go further and check all branches are good for all manifests? We already do that for some devices I guess.

Comment 1

[:gerard-majax]

Currently, only b2g_bumber gets checked against a couple of devices: https://hg.mozilla.org/build/mozharness/file/tip/configs/b2g_bumper/master.py

I am not fluent enough in that mozharness/bumper bot to know if this might be enough to satisfy checking: would a broken base xml break that? None references the sony-aosp-l manifest. The nexus 4 kk and 5 l are there though in https://hg.mozilla.org/build/mozharness/file/tip/configs/b2g_bumper/v2.5.py and we got no warning so I guess this is not enough.

Comment 2

[:gerard-majax]

Attachment: B2G Manifest PR

Comment 3

[:gerard-majax]

Attachment: B2G Manifest PR

Comment 4

[:gerard-majax]

Successfull travis: https://travis-ci.org/mozilla-b2g/b2g-manifest/builds/91406253
Reproducing bug 1224861: https://travis-ci.org/mozilla-b2g/b2g-manifest/builds/91406638

So this is good :)

Comment 5

[Chris AtLee [:catlee]]

Comment on attachment 8687925 [details] [review]
B2G Manifest PR

I suspect this won't work on merges, since $TRAVIS_PULL_REQUEST won't be set.

I think the only way to test would be to set this up on a separate fork in github.

Comment 6

[:gerard-majax]

I'm not sure we can rely on Travis verification to avoid bug 1224861, because according to https://github.com/mozilla-b2g/b2g-manifest/commit/99881044aa5a29ac6a6e9f67d4f962ce28f49630 there was not PR involved to merge that.

Comment 7

[:gerard-majax]

(In reply to Alexandre LISSY :gerard-majax from comment #6)
> I'm not sure we can rely on Travis verification to avoid bug 1224861,
> because according to
> https://github.com/mozilla-b2g/b2g-manifest/commit/
> 99881044aa5a29ac6a6e9f67d4f962ce28f49630 there was not PR involved to merge
> that.

Said otherwise, if we do not run Travis when branching, then we should only care about supporting the PRs, and the changes I made should be enough to limit running that against PR.

Comment 8

[:gerard-majax]

Any other merge happening should have been done from a reviewed PR, so I guess that would be good enough to catch most of the mistakes.

Comment 9

[:gerard-majax]

Should we uplift that to 2.5 ?

Comment 10

[:gerard-majax]

https://github.com/mozilla-b2g/b2g-manifest/commit/2f5bdd132654ae2fcc9ec91ad25520c50976efa7

Comment 11

[:gerard-majax]

Uplifting to v2.5 exposes breakage because of missing base-caf-jb.xml: https://travis-ci.org/mozilla-b2g/b2g-manifest/builds/91652163

Comment 12

[:gerard-majax]

Attachment: B2G Manifest v2.5 PR

Comment 13

[:gerard-majax]

Attachment: B2G Manifest v2.2 PR

Comment 14

[:gerard-majax]

Attachment: B2G Manifest v2.2r PR

Comment 15

[:gerard-majax]

(In reply to Alexandre LISSY :gerard-majax from comment #11)
> Uplifting to v2.5 exposes breakage because of missing base-caf-jb.xml:
> https://travis-ci.org/mozilla-b2g/b2g-manifest/builds/91652163

I have a green build with restoring base-caf-jb.xml: https://travis-ci.org/mozilla-b2g/b2g-manifest/builds/91656235

Comment 16

[:gerard-majax]

So, Chris, it's up to you to decide if we uplift all of that. Fixing v2.5 flame (JB based) can be helpful for contributors that are working on devices stuck on JB (Open C for now for example).