Re: [Qemu-devel] [PATCH 01/10] MAINTAINERS: Take maintainership for QTest

[Paolo Bonzini]

Il 28/04/2014 14:12, Andreas Färber ha scritto:
> code that you refuse to step up as maintainer for (memory,
> Makefile infrastructure, tmp105, ...).

I can step up as maintainer for memory.  At the time I did most of the 
memory work last year, Anthony was still the committer and he was okay 
with community-style maintainership and random pull requests that.

Same for Makefiles, but on the other hand other people such as Peter and 
Michael have done configury work and I don't want to step on their toes.

It looks like we approach the issue from opposite directions.  You want 
as many files as possible covered by formal maintainers.  I don't care 
about that, and prefer more flexibility for long-time contributors.  I 
know Anthony didn't care, and I enjoyed the flexibility.  I learnt that 
*you* care (the tmp105 story).

I don't know how much Peter cares about having stuff covered by 
MAINTAINERS.  The more he does, the more we need to think about how to 
define the status quo in MAINTAINERS.  This means I'd effectively have 
to add myself there in more places.

> So the issues here are twofold,
> a) who sends pulls? Stefan's patches went through me, and no one
> bothered to review libqtest.[hc] API changes of mine. Note that this
> patch does not fix any tree via T:, so it or a variation thereof could
> be applied independently.
> b) who gets CC'ed on and reviews patches? Currently neither me nor
> Stefan nor mst are CC'ed by get_maintainer.pl. This patch adds me.
> Review currently mostly relies on contributors CC'ing undocumented
> people who care to review - extending this patch to a bigger pool of M:s
> is certainly an option.
>
> Generally, our MAINTAINERS file is mixing the concepts of who gets CC'ed
> for review and who actually applies patches; the block layer makes that
> most evident. For qtest we are in need of both aspects.

Ok, this I fully agree with.  Though we need (b) much more than (a).  Do 
you agree?

The problem is that if everything gets a formal maintainer people get 
overloaded.  And I do think that this patch would overload you even 
more, and this is why I don't like it (both for your well-being, and 
egoistically because overloaded maintainers affect contributors 
negatively as well).

What is the meaning of maintainer entries without "T:"?  If it is 
basically "people listed are trusted by Anthony/Peter to send pull 
requests and will shepherd patches into the tree", then Stefan would 
agree to be included for qtest.  It would also a definition that would 
fit what I do for memory, and for Makefiles we could include me+Peter+mst.

Perhaps

Paolo