Re: Clean termination of remote-helpers (was Re: [PATCH 2/2] Add a remote helper to interact with mediawiki (fetch & push))

From: Matthieu Moy <matthieu.moy@grenoble-inp.fr>
Date: Aug 31, 2011, 14:53 UTC
Message-ID: <vpqty8x3a5b.fsf@bauges.imag.fr>
In-Reply-To: <CAGdFq_gSpFm8D1qHs5smUgsqyZXRjw73QFCCkBjTi0n4pwzmHA@mail.gmail.com>

Sverre Rabbelier <srabbelier@gmail.com> writes:

Show 20 quoted lines

> Heya,
>
> On Wed, Aug 31, 2011 at 14:33, Matthieu Moy
> <Matthieu.Moy@grenoble-inp.fr> wrote:
>> I was expecting this part to be more controversial, so I'm just
>> repeating it to draw more attention ;-).
>
> Eek! :)
>
>> # Inform Git that we're done, otherwise Git won't close it's stdin,
>> # and the next loop will be infinite.
>> close(STDOUT);
>> # Flush stdin before we terminate. If we don't, git fetch
>> # (transport-helper.c's sendline function) will try to write to our
>> # stdin, which may be closed, and git fetch will be killed. That's
>> # probably a bug in transport-helper.c, but in the meantime ...
>> while (<STDIN>) {};
>
> Is this caused by you not reading the terminating '\n' that git sends
> when all commands are done?

Indeed. The stream sent by git looks like

import HEAD import refs/heads/master \n <-- this one closes the sequence of import \n <-- this one closes the sequence of commands.

and I was interpreting it as

import HEAD import refs/heads/master \n <-- this one closes the sequence of commands \n <-- what's this??

So it seems the only bug I've found is insufficient documentation. A patch follows.

-- 
Matthieu Moy
http://www-verimag.imag.fr/~moy/