Re: [PATCH] Handle rename of case only, for Windows

From: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Date: Feb 10, 2011, 18:58 UTC
Message-ID: <4D543546.1030207@ramsay1.demon.co.uk>
In-Reply-To: <1296344717.25999.1417928123@webmail.messagingengine.com>

[Sorry for the late reply to this... I'm somewhat backed-up!]

Tim Abell wrote:

>>From ddab003ede9f25d93f4e7eea833523a0aa29d16d Mon Sep 17 00:00:00 2001
> From: Tim Abell <tim@timwise.co.uk>
> Date: Thu, 27 Jan 2011 22:53:31 +0000
> Subject: [PATCH] Handle rename of case only, for Windows

This should not work with the MinGW or msvc build, since they *always* set the st_ino to zero (see do_lstat() and mingw_fstat() in compat/mingw.c). Also, *with default configuration*, it will also not work on cygwin... so saying that this works on Windows may be overstating things a bit ... ;-)

[Hmm, I can't check, but it would probably work on Mac OS X ...]

> Added test to show rename failing in windows.

This fails for me. (er... just to be clear, test #30 fails with your patch applied).

Show 6 quoted lines

> Added test to show that this patch doesn't break protection against overwriting
> an existing file, and another to show that "mv --force" still works.
> 
> Altered mv.c to check whether the source and destination file are actually
> the same file based on inode number before failing.
> If the inode is zero then the data is no use so old behaviour is maintained.

With default configuration, the st_ino will always be zero on cygwin. (see for example the "schizophrenic l/stat() functions" commits adbc0b6, 7faee6b and 7974843, along with the "force core.filemode" commit c869753).

So, you must have core.filemode or core.ignorecygwinfstricks set somewhere in your environment (in the system or global (user) config files?).

[I *always* set core.filemode after git-init or git-clone, so this patch may actually work for me *in practice* (I haven't tested it), but I may well not be a typical user...]

>> Hmm, not so good. st_ino is always 0 on Windows, so this would make
>> false positives, no?
> 
> I tested this on windows 7 under cygwin (which is what I have available) and st_ino reports real numbers for me,

Again, you must have non-default configuration set...

ATB, Ramsay Jones