Re: server.el problem ?

[Eli Zaretskii <eliz@gnu.org>]

> From: Fabrice Popineau <fabrice.popineau@supelec.fr>
> Date: Wed, 1 Aug 2012 14:01:14 +0200
> 
> 
> [1:text/plain Hide]
> 
> Hi all,
> 
> I wanted to use org-protocol with chrome and emacs-24.1 under windows 7.
> I registered the org-protocol protocol with:
> 
> [HKEY_CLASSES_ROOT\org-protocol]
> @="URL:Org Protocol"
> "URL Protocol"=""
> 
> [HKEY_CLASSES_ROOT\org-protocol\shell]
> 
> [HKEY_CLASSES_ROOT\org-protocol\shell\open]
> 
> [HKEY_CLASSES_ROOT\org-protocol\shell\open\command]
> @="\"C:\\Local\\Emacs-24.1\\bin\\emacsclientw.exe\" \"-n\" \"%1\""
> 
> I added a button to the chrome toolbar with :
> 
> javascript:location.href='org-protocol://capture://'+encodeURIComponent(location.href)+'/'+encodeURIComponent(document.title)+'/'+
> encodeURIComponent(window.getSelection())
> 
> but org-protocol failed to capture anything. The reason was that the
> emacsclientw.exe directory was added to the org-protocol url.
> I traced the problem to server.el and I had to patch it this way:
> 
> @@ -1137,7 +1135,8 @@
>                   (let ((file (pop args-left)))
>                     (if coding-system
>                         (setq file (decode-coding-string file coding-system)))
> -                   (setq file (expand-file-name file dir))
> +                   (unless (string-match "^[^/]+:/" file)
> +                     (setq file (expand-file-name file dir)))
>                     (push (cons file filepos) files)
>                     (server-log (format "New file: %s %s"
>                                         file (or filepos "")) proc))
> 
> to prevent addition of the current directory to the org-protocol url.
> This is a crude patch, but anyway, I have the feeling nothing should be
> added to 'file' when it is 'absolute' or when it is a url like in this case.
> 
> Have I done something wrong? Or is there a problem here ?

Are you saying the current code runs a URL through expand-file-name?
If so, I'd say it's sure a bug.  It probably works on Posix platforms
by sheer luck.

Instead of using string-match, perhaps we have a predicate somewhere
that tests strings for being a URL?