> First of all I think we'll have to live with UUID changes. Practically
> speaking these are rarely a problem anyway (our UUID system is broken
> in so many ways that the whole UUID dance is mostly for show)

How can we ensure that none of these UUID changes have add-on/web compatibility 
impact for desktop/mobile?

> We might also want to relax the "builds must have passed green on
> m-i/m-c". People need to be watching their landing after pushing
> anyway and so I don't think it's that different from landing on m-c.

I'm hearing that you'd like devs to be able to simul-land on m-c and m-a. If 
everybody's diligent about backing out when things go awry, I'm OK with that. 
Might get a little hairy when people are landing on top of each other on 
mozilla-aurora, since unraveling test failures from different pushes may be 
more difficult.

-Alex

On Oct 9, 2012, at 9:18 PM, Jonas Sicking <[email protected]> wrote:

> On Mon, Oct 8, 2012 at 3:23 PM, Alex Keybl <[email protected]> wrote:
>> [re-sending, looks like formatting was dropped]
>> 
>> Hi Ehsan,
>> 
>> Just got done discussing this with drivers. Gecko 18 will be the basis for 
>> B2G v1 as things stand today. Here's how things will play out.
>> 
>> 
>> = Landing Process until 11/19, or until told otherwise (mozilla-aurora 18) =
>> * Land on mozilla-aurora (a=blocking-basecamp) if your landing:
>> ** first landed on m-i/m-c and the build/tests are green
>> ** has zero risk to the current desktop/mobile feature set and users (unused 
>> API, ifdef'd out, etc.)
>> ** is blocking-basecamp+
>> ** does not have string/UUID changes
> 
> I'd like to request some changes to these rules in order to make B2G
> development more sane.
> 
> First of all I think we'll have to live with UUID changes. Practically
> speaking these are rarely a problem anyway (our UUID system is broken
> in so many ways that the whole UUID dance is mostly for show). It's
> certainly always possible to avoid UUID changes, but doing so usually
> require a lot of work and also introduces differences between
> mozilla-central and the release branch. Such changes mean risk as well
> as porting pain.
> 
> We might also want to relax the "builds must have passed green on
> m-i/m-c". People need to be watching their landing after pushing
> anyway and so I don't think it's that different from landing on m-c.
> 
> / Jonas

_______________________________________________
dev-b2g mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-b2g

Reply via email to