On Sun, Jan 28, 2018 at 09:30:36PM +0100, Patryk Obara wrote:
> On 28/01/2018 16:57, brian m. carlson wrote:
> >   		if (partial_pack_offset == 0) {
> > -			unsigned char sha1[20];
> > -			git_SHA1_Final(sha1, &old_sha1_ctx);
> > -			if (hashcmp(sha1, partial_pack_sha1) != 0)
> > +			unsigned char hash[GIT_MAX_RAWSZ];
> > +			the_hash_algo->final_fn(hash, &old_hash_ctx);
> > +			if (hashcmp(hash, partial_pack_hash) != 0)
> 
> Maybe "hash" should be struct object_id here?

In this case, I opted not to do that because it's specifically not an
object ID.  It's a checksum for the pack, which isn't a normal Git
object, so I tried to preserve that distinction.

> >   char *index_pack_lockfile(int ip_out)
> >   {
> > -	char packname[46];
> > +	char packname[GIT_MAX_HEXSZ + 6];
> > +	int len = the_hash_algo->hexsz + 6;
> 
> Just me nitpicking, but "len" can be const :)

I wanted it to be const, too, but I recall getting feedback discouraging
it.  I"m happy to make the change; after all, it can only help the
compiler and any future readers.
-- 
brian m. carlson / brian with sandals: Houston, Texas, US
https://www.crustytoothpaste.net/~bmc | My opinion only
OpenPGP: https://keybase.io/bk2204