On Wed, Mar 12, 2014 at 06:22:24PM +0100, Jens Lehmann wrote:
> > Let me know if you still think the hunk should be dropped there.
>
> Yes, I think so. That spot uses memcmp() because ce->name may
> not be 0-terminated. If that assumption isn't correct, it should
> be replaced with a plain strcmp() instead (while also dropping
> the ce_namelen() comparison in the line above). But starts_with()
> points into the wrong direction there.
I think the length-check and memcmp is an optimization[1] here. But we
should be able to encapsulate that pattern and avoid magic numbers
entirely with something like mem_equals(). See my other response for
more details.
-Peff
[1] Getting rid of the magic number entirely means we have to call
strlen(".gitmodules"), which seems like it is working against this
optimization. But I think past experiments have shown that decent
compilers will optimize strlen on a string literal to a constant, so
as long as mem_equals is an inline, it should be equivalent.
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html