Re: convince NTemacs it _can_ symlink

[Eli Zaretskii]

On 20 Mar 2001, Francesco Potorti` wrote:

> Eli Zaretskii:
>    I'm not against the change in principle.  What I'm saying is that such
>    a change should consider all the possible uses of make-symbolic-link,
>    and address them.
> 
> That's next to say that it should never be changed :-)

No, it isn't.  It is to say that I consider this kind of changes to be 
anything but trivial.

> Seriously, though, if programs  test make-symbolic-link to check for the
> ability  to create  symbolic links,  then that's  okay.   Only, symbolic
> links can be made on some file systems, not on others.
> 
> For example, an Emacs running on a unix-like box can make symbolic links
> on unix-like file systems, cannot make them on mounted vfat filesystems,
> can make them on remote ange-ftp file systems if they run on a unix-like
> box,  but not otherwise,  so tramp's  case is  not really  diferent from
> cases we already have at hand.

Tramp's case is not different, but the proposed change doesn't affect 
tramp alone: it affects the whole of Emacs.  In particular, it makes the 
code which have never ran on Windows to be executed on that platform.
All I'm saying is  that, if we care about our users, we should analyze 
the implications of this change and take any preventive actions that follow.

Please understand: I'm not afraid of making Unix-centric code run on 
non-Unix platforms.  FWIW, I  ported many GNU packages to 
MS-DOS/MS-Windows, more than I'd like to remember.  One of those packages 
is GNU Tar, whose port supports hard and symbolic links in the archives, 
inode comparison (on systems which don't know what inode is), file-name
handling (wrt the drive-letter braindamage and backslash nuisance), UID
and GID simulation out of thin air, and whatsnot.  I have enough gray 
hair on my head and in my soul from this and similar experiences to know 
better than to consider such changes as simple ones.

Now, you can take this experience or leave it; I don't call the shots in
Emacs development anyway.  I'm just telling you that whoever thinks this 
is a simple change is mistaken.

I'm not against the change, I'm trying to help those who want it to get 
it right.