[PATCH 1 of 3] Add a new function, filesystem_case

[Paul Moore]

2008/4/23 Matt Mackall <mpm at selenic.com>:
>  First off, a style note: we don't use underbars. I don't ever want to
>  remember whether the function is filesystemcase, filesystem_case,
>  filesystemCase, or FilesystemCase. I always want things in the first
>  style, whether it be a function, class, module, whatever.

Thanks. Patrick has suggested a better name, fspath. I'll go with that.

>  Second, this problem isn't at all localized to Windows, so localizing
>  the solution to Windows is getting off on the wrong foot. Just define
>  the function everywhere and then use checkfolding to decide whether or
>  not to /use/ it.

That sounds fair. I can think of (obscure) cases[1] where this might
be complicated or even outright wrong but it's probably a reasonable
compromise. It may make it harder for a Mac OS expert to later come
along and tweak it to cover normalisation, but I'll worry about that
when it becomes an issue.

[1] A simple example is where finding a guaranteed-foldable filename
could be tricky. In the add command, I want to use the fspath function
on the user supplied filename, which may be 111/222/333 (nothing
foldable there). Of course, in that situation there's no case
sensitivity issue either, so a precheck would do the job.

A more complex (and rare) example is a directory structure a (repo
root, on an ext3 FS), a/.hg, files in a, and a subdirectory a/b which
is a mounted FAT32 volume. Then the case sensitivity varies as we
traverse the path of a file in a/b. Frankly, I'd be tempted to say
that anyone doing that deserves whatever breakage they get...

I'll ponder all this and revise based on it. Thanks for the comments.
Paul.