Re: kill-ring-save and w3m-buffer

[Katsumi Yamaoka <yamaoka@xxxxxxx>]

>>>>> In [emacs-w3m : No.09525] Andreas Röhler wrote:

> GNU Emacs 22.1.50.1 (i686-pc-linux-gnu, X toolkit, Xaw3d scroll bars) of
> 2007-06-07 on karton

That's ok.  (I noticed emacs-w3m running on Emacs 21.x had not
even highlighted the region that is dragged by the mouse if there
are links.)

> This time I probably got (and copied) last version
> correctly.

Unfortunately, you seem to be unsuccessful.

> I'm convinced meanwhile, it's in mouse.el to search the
> reason.

> In line 939 (deactivate-mark) is called and

> form at line 666 relies on that.

Hmm, it's hard to me to see what mouse.el does, but I believe
mouse.el is not a cause of this problem anyway.  IIUC, every
funcall sets the `deactivate-mark' *variable* at the start, and
it leads the `deactivate-mark' *function* to perform at the end.
So, functions that should never deactivate the mark bind the
`deactivate-mark' variable to nil, as it is actually done here
and there in mouse.el.

>> (get-text-property (point) 'help-echo)

> I get
> ... [w3m-track-mouse pos window message nil w3m-url-readable-string
> get-text-property w3m-href-anchor window-buffer] 6]

It should be:
... [w3m-track-mouse deactivate-mark pos window nil ...
                     ^^^^^^^^^^^^^^^
It comes from the `w3m-make-help-echo' macro that I modified
yesterday.  That is *macro*, so functions that use it should be
all recompiled.

Regards,