Re: [PATCH] Fix XEmacs prob & don't require base64 if we don't need to.

[Kai Großjohann]

On Sat, Aug 31, 2002 at 12:26:15PM +1000, Steve Youngs wrote:

> I just built tramp for the first time in ages, came across a problem
> for XEmacs, and I finally got around to addressing an annoyance.
> 
> The problem was that 'find-operation-coding-system' doesn't appear to
> exist in XEmacs.  I had a look through the code and the only place it
> is in Tramp isn't used anywhere (your ChangeLog entry confirms this).
> So I simply wrapped it in a '(unless (featurep 'xemacs)...)'.

What could I use in XEmacs instead?  It's there to work around a bug in
Emacs, eventually, and it would be useful to work around that same bug
in XEmacs, as well.  The Emacs bug is this: Tramp writes a temp file on
the local host, then transfers the temp file to the remote host.  Now
users might have set up Mule variables to choose certain encodings for
certain filenames.  And Tramp should use the REMOTE file name to choose
the encoding, not the name of the local temp file.  So when Tramp passes
the fifth arg VISIT to write-region, that file name should be used for
finding the encoding.

Not sure if XEmacs will do it right anyway.  Do you know?

> The annoyance for me was the '(require base64)'.  Most modern Emacsen
> have the base64 functions built in, but for some reason there are
> always copies of base64.el laying around in people's load-path.  My
> solution here was to wrap the require in a test to see if the
> encode/decode functions are already bound.

Well, Emacsen with base64 implemented in C also provide the `base64'
feature.  IMHO, the right solution would be that XEmacs also provides
that feature.  But as XEmacs does not do that, your patch is the right
workaround, of course.

I'm going to change the order of two lines so that the variable
tramp-feature-write-coding-fix is always defined, only the value will be
wrong in XEmacs.

kai (using mutt from home)
-- 
Life is hard and then you die.