Re: gerrit - does it allow writing commits using a web interface?

[David Kastrup]

Jan Nieuwenhuizen <address@hidden> writes:

> David Kastrup writes:
>
>> We would still need to _track_ patches.  A mailing list is just an
>> unorganized dumping ground.
>
> What exactly do you mean by that, and why can't we do it like linux
> kernel does it?
>
> As I understand it, the submitter keeps reworking and re-posting
> until they get a sign-off and someone puts it in.

More likely until they give up, given our current work force.

> If the submitter loses interest in the patch, is that a problem?

It is a problem if he can't get anybody else interested.

> Nothing keeps us from creating an issue in the tracker, 
> adding a link to the mailing list with the latest patch.

Sure, but if we do that, having a web interface specializing on the
patch management, which is what we hope Gerrit to be capable of, will
make things easier.

> Of course, it would be nice if submitters got lots of positive
> feeback, but I fail to see how a web tool helps with that.

Mostly we at least keep track.

> And, of course, you being the main developer right now, if you
> like the current tools and procedures, that's cool [of course
> you saw Graham's review results and take learning and discouragement
> of git-cl/rietvelt etc into account].

Rietveld sucks because it is a mismatch to git workflows, not because it
is a web interface.

That's why taking a look at more git-focused review tools makes sense.
Dropping the review tools altogether is not feasible: we have a whole
crew working through it and keeping organized, Bug Squad, testers and
Patchy operators, Patch meister and so on.

Those provide important non-programmer infrastructure helping us to keep
our programmers focused.  "Patches floating on a list" won't do the same
for us.

-- 
David Kastrup