Re: Tramp, "[]" in file names, and file-expand-wildcards

[Stefan Monnier]

> > Envelope-to: address@hidden
> > From: "Stefan Monnier" <monnier+gnu/address@hidden>
> > Date: Tue, 09 Oct 2001 13:19:16 -0400
> > 
> > > Maybe I don't understand the issue.
> > > 
> > > Let's take the operation of file-executable-p, for example.  Should
> > > that operation for /cygdrive/d/foo/bar.gz return information about the
> > > compressed or uncompressed file?
> > 
> > Clearly this is a question for jka-compr to answer, so jka-compr
> > needs to be involved (i.e. either the cygwin handler makes sure to
> > pass on the filename (maybe modified) to jka-compr, or jka-compr
> > gets it first).
> 
> But that's the issue, isn't it? the first handler that's called
> depends on the order of package loading into Emacs.  Suppose that
> jka-compr needs to get at the file name first; how can it make sure
> that it will be indeed called first?

Right now, by placing itself carefully in the list of handlers.
But discriminating based on (match-end 0) would work just as well.

By the way, the problem is usually not "to make sure that it gets
there first" but "to get there eventually".  If ange-ftp gets
the filename "/foo:bar/baz.gz" it will pass "bar/baz.gz" to the
ftp client but not to Emacs' file-operations, so jka-compr would
be circumvented.  If ange-ftp were somehow careful to make "bar/baz.gz"
go through the file-name-handlers, then jka-compr would be able to do its
work just fine.


        Stefan