Re: t5570-git-daemon fails with SIGPIPE on OSX

From: Jeff King <peff@peff.net>
Date: Feb 8, 2019, 19:54 UTC
Message-ID: <20190208195459.GA32556@sigill.intra.peff.net>
In-Reply-To: <nycvar.QRO.7.76.6.1902081024550.41@tvgsbejvaqbjf.bet>

On Fri, Feb 08, 2019 at 10:28:12AM +0100, Johannes Schindelin wrote:

Show 23 quoted lines

> Hi Peff,
> 
> On Fri, 8 Feb 2019, Johannes Schindelin wrote:
> 
> > I just had a look at the patch you provided below (for some reason, my
> > previous search on public-inbox only turned up Gábor's mail to which you
> > responded).
> > 
> > Admittedly, I do not really understand all aspects of it, but it applies,
> > still, and I kicked off a stress test here:
> > 
> > 	https://dev.azure.com/git/git/_build/results?buildId=338
> > 
> > It seems that your patch fixes that t5570 flakiness on macOS, and more
> > importantly, addresses an important issue on macOS.
> > 
> > Will play a bit more with it and keep you posted.
> 
> Alas, I was fooled. *Fooled*, I say. Apparently the --stress option makes
> the script *succeed* when it fails?
> [...]
> So I am afraid that your patch does not fix the issue nor does it work
> around it.

I think that patch does the write_or_die conversion to handle EPIPE, but it would still need to turn off SIGPIPE for the whole process.

So you'd also need to stick a:

  sigchain_push(SIGPIPE, SIG_IGN);

somewhere near the start of cmd_fetch(). (There may be a less coarse-grained place to put it, but at this point I think we're just trying to find out whether this approach even solves the problem).

-Peff