Re: [PATCH] Makefile: fix bugs in coccicheck and speed it up

[Jeff King <peff@peff.net>]

On Fri, Mar 05, 2021 at 06:20:47PM +0100, Ævar Arnfjörð Bjarmason wrote:

> >   [after]
> >   $ make coccicheck
> >     SPATCH contrib/coccinelle/free.cocci
> >   make: *** [Makefile:2875: contrib/coccinelle/free.cocci.patch] Error 124
> 
> The "cat $@.log" is back, but there's still some issues with it on
> master now (I didn't change this) it'll spew a lot at you with xargs
> since we emit the whole error output for every file, seemingly, before
> cat-ing the file:
> 
>     make -j1 coccicheck SPATCH_FLAGS=--blahblah V=1 2>&1|grep -i example.*use|wc -l
>     464
> 
> Well, it's a bit better now on my series with a default batch size of 8:
> 
>     $ make -j1 coccicheck SPATCH_FLAGS=--blahblah V=1 2>&1|grep -i example.*use|wc -l
>     88
> 
> I got tired of dealing with the combination of shellscritp and make for
> the day. But maybe something to do as a follow-up if anyone's
> interested.

Hmm, yeah. xargs will keep going unless the command actually returns
255. I never noticed with syntax errors in the cocci files, because
those cause spatch to exit with 255!

IMHO that's likely to be the most common error, though probably "oops,
you don't have spatch installed" is a close second. But that one also
works well, because xargs (at least GNU xargs) will stop if it can't
run the program.

So it might be nice to have it bail on the first failure, but I don't
know that it's worth spending a lot of time on it.

-Peff