bug#21816: elisp-mode-tests fails on a case-preserving filesystem

[Eli Zaretskii <eliz@gnu.org>]

> Cc: Eli Zaretskii <eliz@gnu.org>, 21816@debbugs.gnu.org,
>  Stephen Leake <stephen_leake@stephe-leake.org>
> From: Dmitry Gutov <dgutov@yandex.ru>
> Date: Tue, 3 Nov 2015 18:18:18 +0200
> 
> On 11/03/2015 06:04 PM, Juanma Barranquero wrote:
> 
> > I'm not sure where you're arguing that
> >
> > a) xrefs never need to be compared
> > b) they are sometimes, but it's ok to use equal and fail in
> > case-preserving and case-insensitive systems
> 
> Why don't we canonicalize the file name somehow when an xref instance is 
> created?

Both forms (and some additional ones) are perfectly canonical.

> > I'm a bit surprised to be discussing that the basic type offered by an
> > infrastructure library as xref needs ways to be compared...
> 
> Compared in an impementation-specialized way? Yes, I'd like to see a use 
> case.
> 
> It's very nontrivial to write a comparison function that would work 
> beyond the type you're currently defining (just one xref subtype). 
> Hence, its applicability will probably be rather limited.

This isn't a xref-specific problem, so I think it should have a more
general solution that any package could use.  It's not reasonable to
expect from every Emacs developer to be an expert on subtle aspects of
comparing Windows file names.  We should provide a tested solution for
that.