Re: drag-and-drop failures

[Jan D.]

Dave Love wrote:
> OK, I traced `x-dnd-drop-data'.  The first below is from dragging
> `and' out of a Galeon window running on x86, and the second from the
> same version running on sparc.
>
> Assuming what I see there is what the application is really sending

It is.

> (a) your guess was right; (b) the protocol is completely losing.  I
> guess the only thing Emacs could do about it generally is to try to
> decode as both utf-16be and utf-16le and see which produces least
> U+FFFD, but I doubt that's worth it.  Presumably there should be a
> note about this in PROBLEMS and the Emacs code should swap the bytes
> on big-endian systems, since use between different systems will be
> relatively rare.

I think this is a good approach, I'll implement that (except bytes need not be 
swapped, we just have to use utf-16be instead) shortly.
> Perhaps someone should check whether the behaviour is the same in
> newer versions of Mozilla/Galeon.  I didn't manage to install a recent
> Mozilla on our SPARC/Solaris or Irix systems for reasons I can't
> remember and I can't update the Debian version.

The behaviour is the same.  I was going to file a bug on mozilla on this, but 
there alread is one (http://bugzilla.mozilla.org/show_bug.cgi?id=40608).  I 
added some comment to it.  There has not been much activity on it though.

Note that text/unicode is not an official mime type.  It looks like it was 
invented by mozilla.

Thanks very much for tracking this down.  Still that crash on CDE to fix though.

        Jan D.