git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Bug Report: git add

From
Jeff King <peff@peff.net>
Date
Apr 8, 2011, 19:15 UTC
Message-ID
<20110408191554.GA6516@sigill.intra.peff.net>
In-Reply-To
<7vlizmfrl1.fsf@alter.siamese.dyndns.org>
On Thu, Apr 07, 2011 at 12:28:26AM -0700, Junio C Hamano wrote:
Show 24 quoted lines
> Jeff King <peff@peff.net> writes:
> 
> > @@ -139,8 +138,21 @@ static int lstat_cache_matchlen(struct cache_def *cache,
> >  			if (errno == ENOENT)
> >  				*ret_flags |= FL_NOENT;
> >  		} else if (S_ISDIR(st.st_mode)) {
> > -			last_slash_dir = last_slash;
> > -			continue;
> > +			struct stat junk;
> > +			struct strbuf gitdir = STRBUF_INIT;
> > +			strbuf_add(&gitdir, cache->path, match_len);
> > +			strbuf_addstr(&gitdir, "/.git");
> > +			if (lstat(gitdir.buf, &junk) < 0) {
> > +				if (errno == ENOENT) {
> > +					last_slash_dir = last_slash;
> > +					strbuf_release(&gitdir);
> > +					continue;
> > +				}
> > +				*ret_flags = FL_LSTATERR;
> > +			}
> > +			else
> > +				*ret_flags = FL_GITREPO;
> 
> This only checks "does the directory have .git in it?".

Yeah. I was trying to keep the test as inexpensive as possible, since this is a very frequently called codepath. But really, doing a more elaborate test shouldn't matter. The common case will be that the stat fails, and we do nothing else.

I do worry about adding an extra lstat for each directory having noticeable overhead. Maybe it doesn't matter because of the stat caching, but I didn't measure.

> It probably is sufficient, but setup.c:is_git_directory() may do a more
> appropriate check, I think.  That ".git" thing could be a regular file
> (i.e. "gitdir: path"), so depending on the junk.st_mode, you may have to
> call read_gitfile_gently() on it before checking with is_git_directory().

I worry a little about the PATH_MAX check and die in is_git_directory. I would hate for a deep hierarchy to start failing because of this extra check. OTOH, it is only 5 extra characters to append ".git", so it is unlikely that a path was that close to PATH_MAX but not exceeding it.

Similarly, read_gitfile_gently is anything but gentle. It die()s if we can't open the '.git' file or it is in an invalid format, which would be a regression here.

-Peff
Previous: Junio C HamanoNext: Jeff King
Message 12 of 13 in “Bug Report: git add”
  1. Darren CookApr 5, 2011
  2. Ramkumar RamachandraApr 6, 2011
  3. Jakub NarebskiApr 10, 2011
  4. Ramkumar RamachandraApr 10, 2011
  5. Junio C HamanoApr 11, 2011
  6. Jeff KingApr 11, 2011
  7. Jeff KingApr 7, 2011
  8. Junio C HamanoApr 7, 2011
  9. Jeff KingApr 7, 2011
  10. Jeff KingApr 7, 2011
  11. Junio C HamanoApr 7, 2011
  12. Jeff KingApr 8, 2011
  13. Jeff KingApr 8, 2011

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.