Re: linux-next: ground rules

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Sat, 2009-08-15 at 01:16 +1000, Stephen Rothwell wrote:
> Hi James,
> 
> On Fri, 14 Aug 2009 08:56:13 -0500 James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> >
> > On Fri, 2009-08-14 at 18:02 +1000, Stephen Rothwell wrote:
> > > 
> > >      * successfully unit tested, and 
> > 
> > The rest are fine, but this one isn't feasible for a driver tree ... I
> > don't have all the hardware, and people will insist on fixing
> > theoretical bugs in drivers we can't test on.
> > 
> > A lot of time, bugs turn up in this code only after it has been on
> > release for several months and the small pool of HW owners actually gets
> > around to testing it.
> > 
> > Additionally, I have to carry patches on trust for HW I'm never likely
> > to see outside someones multi-million dollar lab.
> 
> OK, in the context of linux-next, "successfully unit tested" to me means
> that it doesn't break on "reasonable" builds (i.e. x86(_64) allmodconfig
> or something similar) and probably won't break if someone tries to use
> it.  Clearly, you are correct, you can't test everything.  I guess I just
> want to be able to be justifiably annoyed if my builds break for
> something obvious (which does happen from time to time :-().

OK, so I can do compile tested for almost everything except s390
drivers ...

James