Re: [AC] Re: JDE: generated projects not valid Lisp
Vaclav Barta writes:
Steve Youngs wrote:
> APPROVE COMMIT
> |--==> "VB" == Vaclav Barta <vbar(a)comp.cz> writes:
> VB> 2002-03-04 Vaclav Barta <vbar(a)comp.cz>
> VB> * jde.el: use get-file-buffer to map pathname to buffer
> Thanks very much for the patch, Vaclav. I'll apply this shortly.
Er, I've been looking at JDE upstream, and it seems (I didn't test) that
the problem is already fixed in a new beta (2.2.9beta9.1). I do think
the fix is sub-optimal - it uses (the same) lambda in
jde-save-close-buffer and jde-save-variable, where get-file-buffer could
IMHO be used instead - but I'm not sure whether such details are worth
gold-plating... Perhaps there are compatibility problems with
get-file-buffer?
I assume that I used the lambda expression originally because I wasn't
aware of get-file-buffer. I looked up get-file-buffer and found a
reference to find-buffer-visiting, which seems even more suitable
so I am going to try Vasek's patch, but substituting find-buffer-visiting.
Bye
Vasek
P.S.: on the other hand, looking at the jde-save-variable again, how is
it
supposed to work? The fixed (note the comment) lambda is
(lambda (buf)
(let* ((directory-sep-char ?/) ;; XEmacs defaults to backslash
(buf-file-name (buffer-file-name buf)))
(and buf-file-name
(string= (car project)
(expand-file-name buf-file-name)))))
Apparently buffer-file-name depends on the value of directory-sep-char
(not that it's documented); but isn't directory-sep-char above declared
just locally, i.e. a different variable from the global one?
Yes, it shadows the global value allowing me to ensure that for the
purposes of this one function that Emacs/XEmacs uses forward slashes
when returning the path, thus "normalizing the path" for the sake
of comparison.
Paul