On Fri, Feb 17, 2017 at 10:15:31PM -0500, Jeff King wrote:
So for this case, something like the patch below.
Incidentally, there's an off-by-one in the original loop of
stdin_diff_commit that reads past the end of the trailing NUL for the
final sha1 on the line. The problem is the:
pos += GIT_SHA1_HEXSZ + 1;
which assumes we're slurping up the trailing space. This works in
practice because the caller will only permit a string which had a
newline (which it converted into a NUL).
I suspect that function could be more aggressive about complaining about
nonsense on the line, rather than silently ignoring it.
I'd come to basically the same patch, but I did pick up a few niceties
from your patch, like avoiding the off-by-one issue you mentioned above.
Can I place your sign-off on the resulting change?
--
brian m. carlson / brian with sandals: Houston, Texas, US
+1 832 623 2791 | https://www.crustytoothpaste.net/~bmc | My opinion only
OpenPGP: https://keybase.io/bk2204