Re: git's use of mkdir(2)

From: Junio C Hamano <gitster@pobox.com>
Date: Sep 9, 2010, 18:56 UTC
Message-ID: <7v62yen3ts.fsf@alter.siamese.dyndns.org>
In-Reply-To: <201009081936.PAA07078@Sparkle.Rodents-Montreal.ORG>

der Mouse <mouse@Rodents-Montreal.ORG> writes:

Show 7 quoted lines

> I've been trying to convince git to run on some of the systems I use.
> Of particular relevance at the moment are two BSD systems which have an
> important behavioural difference.
>
> Specifically, if /foo/bar does not exist but /foo does,
> mkdir("/foo/bar/",...) works on one and fails showing ENOENT on the
> other.  (Without the trailing slash, it works on both.)

What vintage of BSD do you have that exhibits the problem? It smells like a POSIX violation, considering what "4.12 Pathname Resolution" says on the matter:

  A pathname that contains at least one non-<slash> character and that
  ends with one or more trailing <slash> characters shall not be resolved
  successfully unless the last pathname component before the trailing
  <slash> characters names an existing directory or a directory entry that
  is to be created for a directory immediately after the pathname is
  resolved.

Not that I am saying that such a system does not deserve to be supported, but I am curious to know how widespread the damage is.

> here to ask (a) if this is a known issue and fixed in something more
> recent (I had a look at 1.7.2 and a quick read of the code makes me
> think it still does this, but I could have missed something)

I don't think so---we seem to have a compat/ replacement "mkdir(2)" for MinGW (but that doesn't trim the trailing slash so I would imagine MinGW does not suffer from such a violation), but not for the flavor of BSD you have. It shouldn't be hard to add one, though.