Re: 21.x feature request: windows shortcut support

[Eli Zaretskii]

> From: Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai =?iso-8859-1?q?Gro=DFjohann?=)
> Newsgroups: gnu.emacs.bug
> Date: Thu, 11 Oct 2001 17:05:40 +0200
> > 
> > Apart from the slow-down this would introduce into file-related 
> > operations, I'm afraid this could be tricky, because testing whether a 
> > file exists is done with another primitive that needs to be aware of 
> > shortcuts, and also because you could have a shortcut whose file name is 
> > foo.lnk.lnk.lnk.  You could easily infloop because of these problems.
> 
> Surely the lnk-handler entry needs to be removed from
> file-name-handler-alist before recursively invoking any other
> primitive.

I'm afraid that sometimes the other primitive(s) might need to know
about shortcuts as well.  That would require some elaborate logic
inside each primitive to decide what to do in each case.  Even if it
can be done, it's a mess.

And what if the target of a shortcut is another shortcut?

Look at this from another angle: we are trying to emulate a filesystem
feature here (namely, symlinks).  Since it's a filesystem feature, the
best way to emulate it with minimal complications is as close to the
filesystem level as possible.  Emacs primitives are too high-level for
that; the best place would be one notch before file names are passed
to system calls.