Re: [PATCH 3/5] make "git unpack-file" a built-in
- From
Linus Torvalds <torvalds@linux-foundation.org>
- Date
- Jan 22, 2010, 18:26 UTC
- Message-ID
- <alpine.LFD.2.00.1001221022121.13231@localhost.localdomain>
- In-Reply-To
- <7vmy06vwvy.fsf@alter.siamese.dyndns.org>
On Fri, 22 Jan 2010, Junio C Hamano wrote:
Show 16 quoted lines
> > {
> > unsigned char sha1[20];
> >
> > - git_extract_argv0_path(argv[0]);
> > -
> > if (argc != 2 || !strcmp(argv[1], "-h"))
> > usage("git unpack-file <sha1>");
> > if (get_sha1(argv[1], sha1))
> > die("Not a valid object name %s", argv[1]);
> >
> > - setup_git_directory();
>
> This will now require "git unpack-file -h" to be run in a git controlled
> directory, so strictly speaking it changes behaviour.
>
> Not that anybody would care that much, though.Heh. You didn't notice the same change in the merge-tree conversion that you already accepted ;)
Yeah, any usage messages will now run after setup for the commands that do the whole RUN_SETUP thing. That's pack-redundant, unpack-file, mktag, merge-index and merge-tree. So they'll now report "Not a git repository" before they report invalid arguments.
Linus