End-of-life for b2g-android

[Chris Jones]

We just merged in support for running b2g-gonk on qemu. To my knowledge, that's the last hard blocker for removing b2g-android.

So, b2g-android is now officially unsupported. Huzzah!

This unblocks the following changes, which will start happening in the near future
- removal of JDK/JRE as build tool dependencies
- simplification of INSTALL instructions
- simplification of Makefile
- removal of newly-dead code in glue/gonk; hundreds of MBs of git history. This means faster b2g clones and faster build times.

Cheers,
Chris

----- Original Message -----
> From: "Chris Jones" <cjo...@mozilla.com>
> To: mozilla...@lists.mozilla.org
> Sent: Thursday, December 22, 2011 8:11:55 PM
> Subject: Re: [b2g] Note well: End-of-life for b2g-android coming up
> The end of b2g-android is nigh. Both wifi and 3g-data are close to
> landing. There's one additional dependency that's arisen, which is a
> fallback, non-GL rendering path in b2g-gonk to support the emulator.
> When these pieces are in place, b2g-android is gone!
>
> That will commence the free-for-all to remove unused parts of
> glue/gonk.
>
> Cheers,
> Chris
>
> ----- Original Message -----
> > From: "Chris Jones" <cjo...@mozilla.com>
> > To: mozilla...@lists.mozilla.org
> > Sent: Tuesday, December 13, 2011 11:35:22 PM
> > Subject: [b2g] Note well: End-of-life for b2g-android coming up
> > tl;dr b2g-android will be EOL'd when SMS-on-RIL and wifi or 3g
> > lands.
> > Any objections?
> >
> > Background: currently b2g supports two "backends": the first was
> > b2g-android, which sits on top of Gecko's existing port to Android,
> > including use of Android Java APIs. The second is b2g-gonk, which is
> > a
> > separate porting target from android, and doesn't use any
> > android/java
> > APIs (no dalvik). The b2g-gonk backend is the long-term target for
> > b2g; b2g-android was only "training wheels" for us while we ramped
> > up
> > the b2g-gonk backend.
> >
> > Thanks to amazing work from our team, the b2g-gonk backend is more
> > functional by now than we initially expected :). The performance is
> > already better than b2g-android, and memory consumption is much
> > reduced as well. In some respects b2g-gonk is functional than
> > b2g-android too, including support for outgoing and incoming phone
> > calls. There are still some areas where b2g-android leads b2g-gonk
> > (SMS and wifi/3g data).
> >
> > Instructions for building b2g-gonk are in the INSTALL file in the
> > b2g
> > repo.
> >
> > To consolidate development effort, we'd like to officially "EOL"
> > b2g-android as soon as possible. (Many, maybe most, developers are
> > already on b2g-gonk already.) But we also don't want to delay any
> > ongoing work that depends on b2g-android.
> >
> > I propose EOL'ing b2g-android as soon as the following two APIs are
> > implemented and landed on upstream mozilla-central
> > - SMS on top of RIL (philikon is hacking on this)
> > - wifi or 3g data (mrbkap is hacking on the former, our friends in
> > Spain are hacking on the latter)
> >
> > If anyone has a good reason why we *shouldn't* do this, please let
> > us
> > know asap.
> >
> > Cheers,
> > Chris
> > _______________________________________________
> > dev-b2g mailing list
> > dev...@lists.mozilla.org
> > https://lists.mozilla.org/listinfo/dev-b2g
> _______________________________________________
> dev-b2g mailing list
> dev...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-b2g

[Philipp von Weitershausen]

On Sun, Dec 25, 2011 at 7:19 PM, Chris Jones <cjo...@mozilla.com> wrote:
>  - removal of newly-dead code in glue/gonk; hundreds of MBs of git history.  This means faster b2g clones and faster build times.

OMG it's what I've always wanted for Christmas!