On 12-01-09 09:25 PM, Joel Maher wrote:
> Developers will be writing code for mozilla-central and there is no way around that. We have been keeping stuff in parity for a while between m-c and talos and it works out pretty easy. Basically we find a master source to land on and if something is landing elsewhere we mirror it. Otherwise we do a diff once in a while and get them up to date.
Yup, no matter what we do changes to devicemanager* will have to be
tested against mozilla-central. The question is where stuff should land:
m-c, mozbase, or both.
If we decided mozbase/mozdevice was going to be the canonical location,
workflow for hacking on devicemanager* would probably look like this:
1. Modify devicemanager* in m-c, push to try, make sure things still
work, get review, etc.
2. Apply your patch to devicemanager* to mozbase/mozdevice, push to our
github repo.
3. Sync mozbase/mozdevice in m-c with github repo.
On the other hand, if we are going to make m-c the first place to put
stuff, I guess the workflow looks like this instead:
1. Modify devicemanager* in m-c, push to try, make sure things work, get
review, etc.
2. Push changes to m-c.
We would then just periodically sync the copies of devicemanager* in m-c
with mozbase (much like we were doing before with talos, as you say). I
can see how this is simpler for developers who don't want to have to
care about mozbase, but as Jeff pointed out it is
totally different from everything else we're doing with that project.
That argues against it, but if the pros outweigh the cons, fine (IMO).
The one situation I want to avoid is to have random patches flying into
both repositories on an adhoc basis, which is what just happened with
https://bugzilla.mozilla.org/show_bug.cgi?id=713970.
> I propose m-c is the master source for devicemanager*. This avoid confusion as more people are running tests and hacking from their android builds. If there is a need for major changes in the mozdevice repository I could consider switching. My impression is mozdevice is mostly done and stable.
Yeah, I don't really see a compelling reason to change mozdevice's
interface at this point. Some of the implementation details for the SUT
and ADB variants could probably be better but I think the interface is
about as good as it gets given Android's limitations.
Will