Re: [PATCH] remote-curl: don't hang when a server dies before any output

From: Jeff King <peff@peff.net>
Date: Nov 18, 2016, 17:08 UTC
Message-ID: <20161118170855.22p3vhcocdxgmpy4@sigill.intra.peff.net>
In-Reply-To: <764305554fa74779ad5fb956aa2b658a@EXMBNJE7.ad.twosigma.com>

On Fri, Nov 18, 2016 at 05:04:59PM +0000, David Turner wrote:

Show 8 quoted lines

> > So I don't feel like we have a good patch for the general case yet,
> > and I'm probably not going to get around to implementing it anytime
> > soon. 
> 
> I'm confused -- it sounds like your patch actually does work (that is,
> that Junio's failure was not caused by your patch but by the absence
> of our patches). And your patch handles more cases than mine.  So we
> should probably use it instead of mine.

No, mine passes the vanilla test suite, but fails with GIT_TEST_LONG. If the want/have negotiation takes multiple rounds, the intermediate rounds don't end on a flush packet, and my patch causes remote-curl to complain that the response was truncated.

I think you could fix it by teaching remote-curl that the final packet must be a flush _or_ contain an ACK/NAK, but I didn't try it. That's getting a bit invasive and brittle.

-Peff