Re: [PATCH 7/5] merge-ll: report an error when reading external merge results fails

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 1, 2026, 15:37 UTC
Message-ID: <xmqqpkxt77pd.fsf@gitster.g>
In-Reply-To: <20260930224142.GB763270@coredump.intra.peff.net>

Jeff King <peff@peff.net> writes:

Show 16 quoted lines

> On Wed, Sep 30, 2026 at 11:01:28AM -0700, Junio C Hamano wrote:
>
>> Jeff King <peff@peff.net> writes:
>> 
>> > Here's a resend of that final patch (not just a squash, because the
>> > commit message mentioned the chmod).
>> 
>> Makes sense.
>> 
>> These 6/5 and 7/5 are probably better squashed into 5/5 than left as
>> "oops that was bad, so here is a preliminary clean-up to make the
>> fix easier (6/5), and here is the fix of the fifth step (7/5)", no?
>
> I don't think it is the fault of 5/5 at all (which carefully tried to
> maintain the NULL behavior). The problem fixed by 7/5 existed before my
> series.

Ah, OK, rereading the code before 5/5 is applied, I notice that we are not declaring the result is bad when we jump to "bad:" label after noticing an I/O error. The code only paid attention to the status returned by run_command().

> In theory that fix _could_ come earlier in the series, but it's actually
> much easier to fix after 5/5, because we have a single spot to error
> check.

True.  Thanks.