Re: [PATCH 3/3] cat-file: handle streaming failures consistently

From: Jeff King <peff@peff.net>
To: Junio C Hamano <gitster@pobox.com>
Cc: "Torsten Bögershausen" <tboegi@web.de>,
	"Ævar Arnfjörð Bjarmason" <avarab@gmail.com>,
	"Git Mailing List" <git@vger.kernel.org>,
	"John Szakmeister" <john@szakmeister.net>,
	"Dennis Kaarsemaker" <dennis@kaarsemaker.net>
Subject: Re: [PATCH 3/3] cat-file: handle streaming failures consistently
Date: Wed, 31 Oct 2018 10:37:47 -0400	[thread overview]
Message-ID: <20181031143747.GA9994@sigill.intra.peff.net> (raw)
In-Reply-To: <xmqqva5iurd7.fsf@gitster-ct.c.googlers.com>

On Wed, Oct 31, 2018 at 11:23:48PM +0900, Junio C Hamano wrote:

> Torsten Bögershausen <tboegi@web.de> writes:
> 
> >> +static int stream_blob(const struct object_id *oid)
> >
> > Sorry for nit-picking:
> > could this be renamed into stream_blob_to_stdout() ?
> 
> I think that name makes sense, even though stream_blob() is just
> fine for a fuction that takes a single parameter oid, as there is no
> other sane choice than streaming to the standard output stream the
> blob data.

I was trying to keep the name small since it is a static-local
convenience helper. I'd rather write it as:

  stream_blob(1, oid);

than change the name. ;)

> >> +{
> >> +	if (stream_blob_to_fd(1, oid, NULL, 0))
> >
> > And I wonder if we could make things clearer:
> >  s/1/STDOUT_FILENO/
> 
> What would benefit from symbolic constant more in that function call
> may be CAN_SEEK thing, but s/1/STDOUT_FILENO/ adds negative value to
> that line, I would think.  The name of the function already makes it
> clear this is sending the output to a file descriptor, and an
> integer 1 that specifies a file descriptor cannot mean anything
> other than the standard output stream.

Yes, I'd agree (there are very few cases where I think STDOUT_FILENO
actually increases the readability, since it is usually pretty clear
from the context when something is a descriptor).

-Peff