820594 - generate "engineer/marionette-enabled" b2g builds

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P3)

Description

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

This "new" build type was discovered last week during the transition from Ateam builds to RelEg builds. Its now added to https://etherpad.mozilla.org/b2g-builds. All other build types have been transitioned.

jgriffin: assigning to you first. Please post instructions on how to generate these builds here, and then please reassign this to catlee.

Comment 1

[Jonathan Griffin (:jgriffin)]

The instructions are the same as for existing otoro/nightly builds, except that you omit the VARIANT=user environment variable when running build.sh.  You can explicitly specify VARIANT=eng if you'd like, but it defaults to this, so you can just omit it entirely.

Comment 2

[Jonathan Griffin (:jgriffin)]

s/otoro\/nightly/otoro\/unagi/

Comment 3

[Chris AtLee [:catlee]]

On what frequency should these engineering builds be generated?

On which branches should they be built?

Comment 4

[Jonathan Griffin (:jgriffin)]

To achieve parity with what's on releases.mozilla.com currently, we'd only need builds against mozilla-b2g18 and nightly gaia, once nightly.  However, with the per-commit update testing we plan on standing up with unagi's soon, it would be really helpful to have these builds made per-commit to mozilla-b2g18.

There's no need to generate these builds on other gecko branches, AFAIK.

Comment 5

[Chris AtLee [:catlee]]

Attachment: support --variant for b2g_build.py

This will override the gecko config.

Also, PEP8 FTW.

Comment 6

[Rail Aliiev [:rail]]

Comment on attachment 693001 [details] [diff] [review]
support --variant for b2g_build.py

lgtm

Comment 7

[Chris AtLee [:catlee]]

jgriffin, is there any need to publish manifests for these builds?

Comment 8

[Jonathan Griffin (:jgriffin)]

(In reply to Chris AtLee [:catlee] from comment #7)
> jgriffin, is there any need to publish manifests for these builds?

No there is not.

Comment 9

[Chris AtLee [:catlee]]

Attachment: support --variant for b2g_build.py

I ended up needing a new config file for this since we need a different upload suffix and slightly different set of actions (I'm assuming we don't need to upload manifests for these builds)

Comment 10

[Chris AtLee [:catlee]]

Attachment: set up unagi_eng on mozilla-b2g18

Comment 11

[Rail Aliiev [:rail]]

Comment on attachment 694013 [details] [diff] [review]
support --variant for b2g_build.py

Review of attachment 694013 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good in overall. Just a couple of nits.

::: scripts/b2g_build.py
@@ +273,5 @@
> +        dirs = self.query_abs_dirs()
> +        gecko_config = self.load_gecko_config()
> +        env = self.query_env()
> +        env.update(gecko_config.get('env', {}))
> +        if self.config.get('variant') is not None:

"is not None" can be dropped to improve readability.

@@ +276,5 @@
> +        env.update(gecko_config.get('env', {}))
> +        if self.config.get('variant') is not None:
> +            v = str(self.config['variant'])
> +            env['VARIANT'] = v
> +        if self.config['ccache']:

maybe self.config.get('ccache') to be safer?

Comment 12

[Rail Aliiev [:rail]]

Comment on attachment 694219 [details] [diff] [review]
set up unagi_eng on mozilla-b2g18

Looks like this one is identical to the first patch. A wrong patch maybe?

Comment 13

[Chris AtLee [:catlee]]

Attachment: set up unagi_eng on mozilla-b2g18

oops!

Comment 14

[Aki Sasaki (not active)]

This is in production.