From mboxrd@z Thu Jan  1 00:00:00 1970
From: john at keeping.me.uk (John Keeping)
Date: Sun, 2 Oct 2016 11:45:49 +0100
Subject: [PATCH] cgit: ignore SIGPIPE
In-Reply-To: <CAHmME9r1Jb+3D=8-+V7K9VTNtvTBbTfp1LLvO1fE8fNHgJenpw@mail.gmail.com>
References: <E1bpcTx-00071b-NM@smtp.tt-solutions.com>
 <20161001113300.GQ1673@john.keeping.me.uk>
 <20161001114221.GR1673@john.keeping.me.uk>
 <CAHmME9r1Jb+3D=8-+V7K9VTNtvTBbTfp1LLvO1fE8fNHgJenpw@mail.gmail.com>
Message-ID: <20161002104549.GA1997@john.keeping.me.uk>

On Sat, Oct 01, 2016 at 10:52:11PM +0200, Jason A. Donenfeld wrote:
> I see the utility of this with something like a highlight filter gone
> bad. But is it safe to do this in the context of an authentication
> filter? What's the failure behavior like once this patch is applied?

If we ignore SIGPIPE then we get an EPIPE error from write(2); in both
places that use write(2) we die if this happens.

So I think the only changes are our exit code and the fact that with
this patch we print an error message rather than exiting silently.