Hi Darren,

On Sat, 23 Feb 2019 09:52:07 -0800 Darren Hart <dvhart@infradead.org> wrote:
>
> Apologies if I've asked you this before... I didn't find it after some
> searching.
> 
> We should be catching errors like this before they hit next. First,
> there is no reason we can't catch them - unlike the integration failures
> only next can catch. Second, once they are in next, there is no "right"
> way to fix them.  Either rebase or send the bad patch to mainline - both
> are bad. Don't get me wrong, I'm glad next catches them.... but I'd like
> to get to the point where it doesn't trigger on our subsystem :-)
> 
> Are your patch mechanics tests available for us to integrate into our
> commit and prepublication checks?

I have attached the current version of my check scripts (I can't use
git hooks since I want to do these checks when I fetch trees).  They
both take a range of commits (I usually pass "^origin/master <old SHA
for branch>...<new SHA for branch>" (only 2 dots if the branch is just
fast forwarded).

These could obviously be simplified because they also generate most of
the appropriate mail messages ...
-- 
Cheers,
Stephen Rothwell