[coreboot] Patch management

ron minnich rminnich at gmail.com
Sun May 31 23:14:37 CEST 2009


On Sun, May 31, 2009 at 4:41 PM, Peter Stuge <peter at stuge.se> wrote:
> ron minnich wrote:
>> > I would like to request a better patch management system than this
>> > mailing list.
>>
>> it's a hard problem.
> ..
>> All in all, I think the process works.
>
> I don't, when patches are untracked and linger for however long until
> the author or someone else sends a ping, at which point they may get
> some more attention, or will just continue to linger.

all patches or some patches? Every time I svn, which is almost daily,
there are changes.

Clearly, some patches are being reviewed and acked.

>
> Especially when there is no feedback,


OK, I'm guilty here. I don't review every patch. Does anyone? Could anyone?

There is some bit of burden here on people who submit patches. A patch
needs to be very clear as to:
- what is the issue
- how is the issue solved
- why is it solved in this way

We do get patches from time to time that are confusing. I see it
happen. We get a patch, for example, which solves a problem that
nobody understood to be a problem, and that is not clear as to how it
solves the problem. It is sometimes unclear that the patch is a good
idea. At the same time, it's hard to take the time to write up just
why (in my case) I'm not comfortable with or don't understand the
patch.

No code review system, however automated, can fix that kind of problem.

Again, I see this problem on other projects. On these other projects I've seen:
- patches that fix a design problem that people aren't willing to
admit is a design problem
- patches that fix an implementation problem -- same story, it's hard
to get people to get it
and at some point you give up
- patches that add a very nice capability -- people argue about silly
things and avoid the main
point
- patches that fix a problem -- badly -- such that nobody ever takes
it in, and it dies a
lingering, but well-deserved death.

I've seen all these scenarios on coreboot. I've submitted bad patches
that got NAK'ed or died. I've submitted stuff I thought we ought to
have, same experience. I've wrongly ACK'ed and NAK'ed patches.

It happens. We're fallible. But again, v2 rev level is 4224 or so,
which means that in some way, the system worked at least 4000 times.

All that said, if somebody wants to try out and evangelize one of
these improved systems, and it can improve our life, why not try it?
Computers are there to remove drudgery from our lives and smooth our
path for us -- let's use them :-)

thanks

ron




More information about the coreboot mailing list