Re: [bug report] git-am applying maildir patches in reverse

From: Jeff King <peff@peff.net>
Date: Mar 3, 2013, 06:32 UTC
Message-ID: <20130303063244.GB23221@sigill.intra.peff.net>
In-Reply-To: <7vtxotc895.fsf@alter.siamese.dyndns.org>

On Sat, Mar 02, 2013 at 10:22:46PM -0800, Junio C Hamano wrote:

Show 9 quoted lines

> Andreas Schwab <schwab@linux-m68k.org> writes:
> 
> > You should always cast to unsigned char when determining the order of
> > characters, to be consistent with strcmp/memcmp.
> 
> We treat runs of digits as numbers, so it is not even similar to
> strcmp.  As long as it is internally consistent (i.e. the return
> value inside the loop (*a - *b) must match the last return), it
> should be OK, no?

I almost responded and said something similar, but we also do byte-wise comparisons for non-numeric elements, and we would want those to match what other programs may do (and what git used to do).

I highly doubt that it matters in practice, as it would mean:

  1. The sorting of a maildir's filenames are dependent on the sorting
     of non-numeric bits. We can't rule out such a scheme, but I'd guess
     implementations either use numbers, or their sort order is
     meaningless (and that is what I found in the ones I looked at).

  2. The importantly-sorted bits contain non-ascii characters (the
     difference is only seen when we go outside the signed range).

but it doesn't hurt to be thorough (and to set a good example).

-Peff