Re: surprising error message in parse_opt_with_commit

From: Tim Harper <hidden>
Date: 2016-06-15 22:47:11

On Thu, Aug 6, 2009 at 1:34 PM, Shawn O. Pearce[off-list ref] wrote:
Tim Harper [off-list ref] wrote:
quoted
 610 int parse_opt_with_commit(const struct option *opt, const char
*arg, int unset)
 611 {
 612  unsigned char sha1[20];
 613  struct commit *commit;
 614
 615  if (!arg)
 616          return -1;
 617  if (get_sha1(arg, sha1))
 618          return error("malformed object name %s", arg);
 619  commit = lookup_commit_reference(sha1);
 620  if (!commit)
 621          return error("no such commit %s", arg);
 622  commit_list_insert(commit, opt->value);
 623  return 0;
 624 }

It appears the get_sha1 call is returning true, causing the 'malformed
object name' error to be returned.  However, it seems that ideally
since efabdfb is not malformed (it would be a valid ref if it
existed), the execution path should continue to line 619, receive no
commit, and fail on 621.
get_sha1 is responsible for expanding an abbreviated ID to the
full ID.  If it can't do the expansion, it errors out.  The code
is correct as-is, though the error message on 618 is a bit odd.

--
Shawn.
ok, that makes more sense.  I notice if I pass a full, mutated
40-character sha1 commit, I get the error message I expect.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help