[PATCH 1/2] Effective Maintainer Work Flow

Andy Green andy at openmoko.com
Mon Jan 21 10:40:09 CET 2008

Hash: SHA1

Somebody in the thread at some point said:

> Much nicer :-) I wish we could do this with fewer items scattered
> all over various structures, though. I've simplified the irq_works
> logic a bit, mainly to make sure we always initialize the field.

Actually Werner, this was about removing your comments.  I did point out
to you that the patch you took from git wasn't finished.

> Please check if I got everything right. It seems that your patch was
> made against an old code base, so there were a number of conflicts I
> had to fix. I hope I found all of them, but you never know ...

What old code base?  It was up to date with the svn patchset at the
time, as are both the kernel and u-boot git as of yesterday with
yesterday's set and upstream.  If it won't merge, tell the submitter on
the list what versions you are using and ask for resubmission.

I think we need to change the way we are working to conform with how
other projects handle merging.  If you have an objection to a patch,
especially a big new feature, NAK it on the list explaining your
objections and let the submitter decide to accept your views or argue
them, either way it's up to the *submitter* to make the changes and
issue Try#<n+1>.  Your shoving the submitter away from their keyboard
and making the changes yourself, then committing it without further
comment from anyone else doesn't encourage submission.

Certainly your current method of trying to "own" everything going
through your hands will not scale, and in fact will block you working on
anything else: nor did anybody die and make you the one person that can
randomly meddle with code and commit it without the same procedure as
everyone else.

Let's not pretend that by glancing at the patch you understood the
intricacies of MCI and the platform interrupts (which you broke with
your earlier meddling) or whatever other unfamiliar subsystem it might
be, instead make your request to the guy whose head is full of the
detail already.  You needed to expend a large effort to understand MCI
because a patch came where the submitter understood it, you need to
fully grok everything anybody sends a patch for?  Nope.  What happens if
people rain down large patches on stuff you aren't familiar with?  You
just need to filter and apply patches, a far lighter problem.

That said, your *review* is appreciated, but that's all this should have
been pending a Try #2 from the *submitter*, which is what I would see on
the other projects I spent time with.

- -Andy

Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org


More information about the openmoko-kernel mailing list