Like I said on IRC, I saw that git describe --contains has a bad
behaviour:
$ git describe --match='asd*' HEAD; echo $?
fatal: cannot describe 'e272415ab7da3bde51af2ce95c88d7be3abfba28'
128
$ git describe --contains HEAD; echo $?
undefined
0
THe "undefined" output is on stdout (not stderr), and returns 0.
The issue here is that it internally uses git-name-rev by exec-ing it, which
makes it hard to fix. Though I suppose that we could instead of fork-ing
share some logic with builtin-name-rev.c, but I'm not at home yet, so
won't likely have a patch for this issue.
Note that the use of the "new" --match here was just to be unable to
describe the HEAD to show the difference, the inconsistency has nothing
to do with the patch I propose, I just happen to noticed that.
AFAICT it's not a regression, it's just a misfeature :)
--
·O· Pierre Habouzit
··O madcoder@xxxxxxxxxx
OOO http://www.madism.org
Attachment:
pgpW4YOBavH8p.pgp
Description: PGP signature
