[Date Prev][Date Next][Thread Prev][][Date Index][Thread Index]

Re: ftp oddities

Katsumi Yamaoka <yamaoka@jpl.org> wrote:
> I see.  The problem is that emacs-w3m uses ange-ftp even if
> w3m-use-ange-ftp is nil.  What ever on earth has happened?!

Since I built emacs-w3m from cvs, would it be possible that I need to
turn on a particular option or include a certain library to enable
emacs-w3m native ftp support?

> Try the following and let us know the result, that is, the Lisp
> backtrace. (Copy it into the *scratch* buffer and type `C-j' at
> the end of the last line.)
> (progn
>   (setq debug-on-error t)
>   (load "w3m.el")
>   (fset 'w3m-goto-ftp-url (lambda (&rest args)
> 			    (error "!! `w3m-goto-ftp-url' is called")))
>   (unwind-protect
>       (w3m-goto-url "ftp://ftp.gwdg.de/";)
>     (load "w3m")))

When I executed this code (w3m-use-ange-ftp nil, no ange-ftp
customisation), the root directory of the server was displayed
correctly and I heard the beep as described before. Once I wanted to
enter one of the subdirectories, the debug buffer popped up and showed
the following:
Debugger entered--Lisp error: (ftp-error "FTP Error: DIR failed: 425 Can't build data connection: Connection refused.")
  signal(ftp-error ("FTP Error: DIR failed: 425 Can't build data connection: Connection refused."))
  ange-ftp-error("ftp.gwdg.de" "anonymous" "DIR failed: 425 Can't build data connection: Connection refused.")
  ange-ftp-ls("/anonymous@ftp.gwdg.de:/" "-al" t nil)
  apply(ange-ftp-file-exists-p "/anonymous@ftp.gwdg.de:/spinner.gif")
  ange-ftp-hook-function(file-exists-p "/anonymous@ftp.gwdg.de:/spinner.gif")
  w3m-goto-url("ftp://ftp.gwdg.de/pub/"; nil nil nil "ftp://ftp.gwdg.de/"; ((lambda (&rest --cl-rest--) (apply ... ... ... ... ... --cl-rest--))))
  byte-code("Æ	lj\n&\fpLˆ
È Lˆ‡" [url reload w3m-current-url handler --newbuffer--70121 --wconfig--70122 w3m-goto-url nil current-window-configuration] 7)
  w3m-view-this-url(nil nil)

> If the "!! `w3m-goto-ftp-url' is called" message doesn't appear in
> the beginning of the *Backtrace* buffer, there might be another
> cause.
Apparently, that's precisely the case.