Re: 5088/3: the SPI dependency

[Russell King - ARM Linux <linux@xxxxxxxxxxxxxxxx>]

On Mon, Jul 21, 2008 at 01:59:13PM -0700, Andrew Morton wrote:
> On Mon, 21 Jul 2008 10:58:50 +0100
> Russell King - ARM Linux <linux@xxxxxxxxxxxxxxxx> wrote:
> 
> > Since we are over a week into the merge window, and I see no sign of
> > the SPI tree merging, it's time to make a decision on what to do about
> > the outstanding changes in my tree which depend on the SPI tree having
> > been merged first (iow, the presence of linux/spi/max7301.h.)
> 
> gpio-gpio-driver-for-max7301-spi-gpio-expander.patch is in -mm, and -mm
> merges after the subsystem trees, usually in the second week of the
> merge window.  It has been this way for many years.

My message, which you've taken to be something of an insult, was to find
out what to do with all the other peoples work in my tree to allow things
to progress.

> Well you could have just asked me to merge that patch early, as I
> regularly do when people ask me to.  I knew nothing about any of this,
> so of course that patch has been following the usual procedure.

Well, I've ONLY JUST found out that it's you I'm waiting for.  Previously
to this, I thought I was waiting for the SPI tree to merge.

Right.  Clearly there's a complete lack of communication by others here
to keep me informed about what their patches depend on, who they depend
on, and what's being done to satisfy those dependencies.  Specifically:

1. I was not informed that this patch relied upon the max7301 SPI
   support, until I found that it didn't build.

2. You were not informed that the max7301 SPI support was required for
   patches in the ARM tree.

The question is what is to be done about it?

Obviously, I can not do (2) until I know that (a) there is a dependency
issue and (b) you have the necessary patch and (c) you're not already
aware of that.

Clearly, the right person to ensure that everyone knows about the
dependency they're creating is the one submitting the original patch.
They shouldn't leave it up to people upstream to guess or discover their
patch dependencies for them.

This all makes me think that accepting patches which have such
dependencies is a Very Bad Idea, and given your reaction to my message
about this, I really don't want to accept these kind of patches ever
again.

-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:        http://www.arm.linux.org.uk/mailinglists/faq.php
Etiquette:  http://www.arm.linux.org.uk/mailinglists/etiquette.php