Jeffrey Juliano writes:
> Pete Forman wrote:
> [...]
> > Any suggestions for syntax? My first thought is to use
> > semicolons instead of colons. Something like a URI would have
> > been nice but efs would munch that up as well.
>
> I agree with Kai's idea that URL type names would be good.
>
> However, if you have to change to non-colons for now, that would
> also solve the problem I'm seeing on NT (where colon has one of two
> special meanings in a pathname, depending on where it is in the
> pathname)
I've customized the three relevant variables and that seems to be
working.
> Is this a general design flaw in the filehandling code in emacs?
> If so, do you think we could make a case to the powers-that-be that
> it would be really useful to us (and possibly others in the future)
> to fix it?
>
> (have to admit, I don't fully grok the cause of the issue you're
> seeing with EFS)
My theory is
emacs file handling is or was flawed.
efs does some nasty low level stuff to get round the problems.
The ideal fix would be
emacs is improved to exploit file-name-handler-alist better, and
to provide more hooks in file name handling.
efs should then remove its overloading of expand-file-name, load
and require, and stick to lisp variables and hooks.
A more radical idea would be to get ange-ftp/efs to drop their current
syntax altogether and adopt a URI style name, i.e.
ftp://user:password@host:port/cwd1/.../cwdN/name;type=typecode
rcp.el would then introduce schemes like rcp and scp instead of ftp.
At some stage we might submit these to the IETF. Note that telnet is
already in RFC 1738, though just in the context of an interactive
session. It should be possible to extend it to represent a file.
While we're at it, why should file names have to start with '/'? That
should just mean "absolute path on local filesystem". Or in URI
parlance, it should be a synonym for "file:///".
There are some problems in URIs which would require attention. A
couple that spring to mind are the lack of relative URIs in the file
scheme, and the cruftiness of absolute ftp names.
--
Pete Forman
Western Geophysical
[EMAIL PROTECTED]
