[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: still no go with tramp
Re: still no go with tramp
Sun, 4 Jan 2004 20:07:26 +1100 (EST)
On Sat, 3 Jan 2004, Kai Grossjohann wrote:
> Tim Connors <address@hidden> writes:
> > On Sat, 3 Jan 2004, Kai Grossjohann wrote:
> >> How come that the find-file command says
> >> "/address@hidden/home/office/tconnors/magellanic/work/flag", but
> >> the error says "/[rsyncfsh/"?
> >> This must have happened somewhere, but where?
> > I've got
> > '(tramp-default-method "rsyncfsh")
> > Does that make sense?
> That might explain a transformation from
> "/address@hidden/home/office/tconnors/magellanic/work/flag" to
> but not to "/[rsyncfsh/".
> Kai, still confused
I'm not entirely sure it's tramp's fault. What always made me confused was
I saw the temp file sitting there in /tmp for a few second - so tramp
seems to be working. But that error message certainly has the wrong info.
I finally got edebug to work (It wouldn't let me instrument
(find-file-noselect) until I explicitly loaded the edebug.el file)
For your reference, I'll include the a large part of the defun, but the
relevant part is highlighted down the bottom:
(defun find-file-noselect (filename &optional nowarn rawfile)
;;;.... lots before here
;; Else: we must create a new buffer for filename
;;; The truename stuff makes this obsolete.
;;; (let* ((link-name (car (file-attributes filename)))
;;; (linked-buf (and (stringp link-name)
;;; (get-file-buffer link-name))))
;;; (if (bufferp linked-buf)
;;; (message "Symbolic link to file in buffer %s"
;;; (buffer-name linked-buf))))
(setq buf (create-file-buffer filename))
;; Catch various signals, such as QUIT, and kill the buffer
;; in that case.
(insert-file-contents-literally filename t)
(insert-file-contents filename t))
(when (and (file-exists-p filename)
(not (file-readable-p filename)))
(signal 'file-error (list "File is not readable"
;; Unconditionally set error
(setq error t)
;; Run find-file-not-found-hooks until one returns
;; If they fail too, set error.
(setq error t)))))
;; Find the file's truename, and maybe use that as visited
;; automatically computed in XEmacs, unless jka-compr was
(setq buffer-file-truename truename))
(setq buffer-file-number number)
;; This should be in C. Put pathname
;; abbreviations that have been explicitly
;; requested back into the pathname. Most
;; importantly, strip out automounter /tmp_mnt
;; directories so that auto-save will work
(setq buffer-file-name (abbreviate-file-name
;; Set buffer's default directory to that of the file.
(setq default-directory (file-name-directory
;; Turn off backup files for certain file names. Since
;; this is a permanent local, the major mode won't
(and (not (funcall backup-enable-predicate
(setq backup-inhibited t)))
;;;; starting here
;; #### FSF 20.3 sets buffer-file-coding-system to
;; `no-conversion' here. Should we copy? It also
;; makes `find-file-literally' a local variable
;; and sets it to t.
(after-find-file error (not nowarn))
(setq buf (current-buffer))))
;;;; ending here
(signal (car data) (cdr data))))
I trace the function, and the error is signalled in rawfile
test. If I just comment this out, the file loads proerply, the format is
raw. If left in, the error is signalled, the buffer deleted, and I don't
get my file.
Now, I recall that mule causes all sorts of complications with file
encodings, but I have xemacs21-nomule installed. So just to test this out
- I installed xemacs21-mule. Same probs.
Any idea Kai, or should I forward this to xemacs-beta?
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
"Whenever someone thinks that they can replace SSL/SSH with something much
better that they designed this morning over coffee, their computer
speakers should generate some sort of penis-shaped sound wave and plunge
it repeatedly into their skulls until they achieve enlightenment."
--Peter Gutmann/Matt Robinson