On Tue, Oct 30, 2012 at 07:46:31AM +1100, Stephen Rothwell wrote:

> Then you need to have a different branch being tested by Fenguang and
> friends to what you have in linux-next.   The stuff that you put into
> linux-next has to be at least build tested, so you should delay pushing
> it up to your linux-next included branch until after that is done.

Which is what I was doing, as I said I had thought I'd pushed up
something different to what was actually there.

> This case was particularly irritating because you were the developer of
> the patch that was broken which indicates that you are not even testing
> your own development work before pushing it out for others to work on top
> of.

I don't treat my own patches any differently to any other patches I push
out; in this case the issue was a mistaken fix for a merge issue cherry
picking the code out of a working branch into the public branch.  

With a huge chunk of new device enablement like this it's just not
possible to do anything except build testing on the branches in -next,
there'll be at least some cross tree issues getting in the way of
actually running anything.  At that point all the issues with time spent
doing build tests apply; as I keep saying I will tend to do some if I
have time but it can fall by the wayside and they're not the tests
anyone else does.