Re: Patch to change mouse binding for html view in gnus

Re: Patch to change mouse binding for html view in gnus

[Reiner Steib]

[ Cc-ing the Gnus list; full quote. ]

On Sun, Jul 01 2007, Tom Tromey wrote:

> I have w3m installed, and so when Gnus sees an HTML message it is
> displayed via w3m.  This is quite nice.
>
> However, if I click on a link in an HTML message rendered this way,
> the target is displayed using w3m.  I generally don't want this; I
> only use w3m in Emacs for specialized things and in other cases I'd
> rather use my external web browser.

I'm not sure if doing this unconditionally is a good idea.  Shouldn't
it be customizable?

> This patch changes Gnus to use browse-url rather than w3m for links in
> buffers it renders.
>
> Tom
>
> 2007-07-01  Tom Tromey  <tromey@redhat.com>
>
> 	* mm-view.el (mm-inline-text-html-render-with-w3m): Locally set
> 	w3m-goto-article-function.
>
> cvs diff: Diffing .
> Index: mm-view.el
> ===================================================================
> RCS file: /sources/emacs/emacs/lisp/gnus/mm-view.el,v
> retrieving revision 1.29
> diff -u -r1.29 mm-view.el
> --- mm-view.el	19 Apr 2007 12:08:53 -0000	1.29
> +++ mm-view.el	1 Jul 2007 20:58:45 -0000
> @@ -252,6 +252,7 @@
>  	(let ((w3m-safe-url-regexp mm-w3m-safe-url-regexp)
>  	      w3m-force-redisplay)
>  	  (w3m-region (point-min) (point-max) nil charset))
> +	(set (make-local-variable 'w3m-goto-article-function) #'browse-url)
>  	(when (and mm-inline-text-html-with-w3m-keymap
>  		   (boundp 'w3m-minor-mode-map)
>  		   w3m-minor-mode-map)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: Patch to change mouse binding for html view in gnus

[Tom Tromey]

>>>>> "Reiner" == Reiner Steib <reinersteib+gmane@imap.cc> writes:

>> However, if I click on a link in an HTML message rendered this way,
>> the target is displayed using w3m.  I generally don't want this; I
>> only use w3m in Emacs for specialized things and in other cases I'd
>> rather use my external web browser.

Reiner> I'm not sure if doing this unconditionally is a good idea.  Shouldn't
Reiner> it be customizable?

I don't know.  Personally I think this is a good default -- people who
want to browse extensively in Emacs can set their
browse-url-browser-function.

Maybe there are people who want to usually use an external browser but
then browse using w3m from gnus.  That seems weird to me :), but if
such people exist then, yeah, this should be customizable.

Tom

Re: Patch to change mouse binding for html view in gnus

[Katsumi Yamaoka]

>>>>> In <m3r6nrajgr.fsf@fleche.redhat.com> Tom Tromey wrote:
>>>>>> "Reiner" == Reiner Steib <reinersteib+gmane@imap.cc> writes:

>>> However, if I click on a link in an HTML message rendered this way,
>>> the target is displayed using w3m.  I generally don't want this; I
>>> only use w3m in Emacs for specialized things and in other cases I'd
>>> rather use my external web browser.

Reiner> I'm not sure if doing this unconditionally is a good idea.  Shouldn't
Reiner> it be customizable?

> I don't know.  Personally I think this is a good default -- people who
> want to browse extensively in Emacs can set their
> browse-url-browser-function.

> Maybe there are people who want to usually use an external browser but
> then browse using w3m from gnus.  That seems weird to me :), but if
> such people exist then, yeah, this should be customizable.

Isn't it sufficient to use the following?

(setq w3m-goto-article-function #'browse-url)

This is used only when the `w3m-safe-view-this-url' command is
invoked, and normally the `w3m-safe-view-this-url' command is
bound to a certain key only in html articles.

Re: Patch to change mouse binding for html view in gnus

[Tom Tromey]

>>>>> "Katsumi" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

Katsumi> Isn't it sufficient to use the following?
Katsumi> (setq w3m-goto-article-function #'browse-url)

Katsumi> This is used only when the `w3m-safe-view-this-url' command is
Katsumi> invoked, and normally the `w3m-safe-view-this-url' command is
Katsumi> bound to a certain key only in html articles.

I'm afraid I'm not a w3m expert (or a gnus expert for that matter :-).
If this is what you would prefer I am happy to send a new patch.
Just let me know.

thanks,
Tom

Re: Patch to change mouse binding for html view in gnus

[Katsumi Yamaoka]

>>>>> In <m38x9y90k1.fsf@fleche.redhat.com> Tom Tromey wrote:
>>>>>> "Katsumi" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

Katsumi> Isn't it sufficient to use the following?
Katsumi> (setq w3m-goto-article-function #'browse-url)

> I'm afraid I'm not a w3m expert (or a gnus expert for that matter :-).
> If this is what you would prefer I am happy to send a new patch.
> Just let me know.

I recalled I wrote the same suggestion to you last month. ;-)
See: http://news.gmane.org/group/gmane.emacs.w3m/thread=6852

TSUCHIYA Masatoshi implemented this variable in Jan 2004 for the
other purpose (to bind it to a certain Lisp function in
nnshimbun.el[1]), but it is just the thing for people like you.

Though I don't represent a majority (because I am a member of
the emacs-w3m team), I am satisfied with the nil value of this
variable.  I've set `w3m-make-new-session' to t in order to make
emacs-w3m pop a new tab up when Gnus (or other) requires to
visit a web page.  There is no much case for which I want to use
an external browser.  In such a case, I will type `M' in the
emacs-w3m buffer.  Anyway, I don't think it is necessary to set
it to a certain value in Gnus.

[1] http://article.gmane.org/gmane.emacs.w3m/3797

Re: Patch to change mouse binding for html view in gnus

[Tom Tromey]

>>>>> "Katsumi" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> I'm afraid I'm not a w3m expert (or a gnus expert for that matter :-).
>> If this is what you would prefer I am happy to send a new patch.
>> Just let me know.

Katsumi> I recalled I wrote the same suggestion to you last month. ;-)
Katsumi> See: http://news.gmane.org/group/gmane.emacs.w3m/thread=6852

Sorry, I misunderstood that message.  I didn't realize that
w3m-goto-article-function was solely for this purpose -- I thought it
was more generic, and thus needed to be set buffer-local.

Katsumi> Though I don't represent a majority (because I am a member of
Katsumi> the emacs-w3m team), I am satisfied with the nil value of this
Katsumi> variable.  I've set `w3m-make-new-session' to t in order to make
Katsumi> emacs-w3m pop a new tab up when Gnus (or other) requires to
Katsumi> visit a web page.  There is no much case for which I want to use
Katsumi> an external browser.  In such a case, I will type `M' in the
Katsumi> emacs-w3m buffer.  Anyway, I don't think it is necessary to set
Katsumi> it to a certain value in Gnus.

The reason I think this patch is needed is that I've already
configured Emacs to visit URLs the way I like.  I was surprised to
find that I had to make another modification, in a different and more
obscure (w3m-goto-article-function is not customizable) place to get
the effect I want.

If you do all your browsing in Emacs then you probably already have
browse-url-browser-function set as you like...

Tom

Re: Patch to change mouse binding for html view in gnus

[Katsumi Yamaoka]

>>>>> In <m3fy454d7y.fsf@fleche.redhat.com> Tom Tromey wrote:

Katsumi> I don't think it is necessary to set it to a certain value in
Katsumi> Gnus.

> The reason I think this patch is needed is that I've already
> configured Emacs to visit URLs the way I like.  I was surprised to
> find that I had to make another modification, in a different and more
> obscure (w3m-goto-article-function is not customizable) place to get
> the effect I want.

I noticed `w3m-goto-article-function' hasn't been made up as a
user option, so I've made it customizable in the emacs-w3m CVS
trunk yesterday.  Feel free to set it to a Lisp function you
like, but I noticed there is a limitation a bit (because it was
introduced for a particular purpose first).  That is the function
set should return a non-nil value.  Though `browse-url' seems to
return a non-nil value normally, it is better to ensure it.  For
this reason, I recommend using the function form

(lambda (url) (browse-url url) t)

rather than just using `browse-url'.

> If you do all your browsing in Emacs then you probably already have
> browse-url-browser-function set as you like...

I leave all those as the default and have no problem (Firefox
happens to be used by default).  I don't always necessarily use
emacs-w3m.   I use Firefox when purchasing something, manipulating
my bank account... ;-)

Re: Patch to change mouse binding for html view in gnus

[Katsumi Yamaoka]

>>>>> In <b4md4z9krec.fsf@jpl.org> Katsumi Yamaoka wrote:
>>>>>> In <m3fy454d7y.fsf@fleche.redhat.com> Tom Tromey wrote:

> but I noticed there is a limitation a bit (because it was
> introduced for a particular purpose first).  That is the function
> set should return a non-nil value.  Though `browse-url' seems to
> return a non-nil value normally, it is better to ensure it.  For
> this reason, I recommend using the function form

> (lambda (url) (browse-url url) t)

> rather than just using `browse-url'.

I've improved it so that just `browse-url' may work no matter
what it returns (in the emacs-w3m CVS trunk).  So, you may use
the one that I suggested first.

>>>>> In <b4mbqevzf0y.fsf@jpl.org> Katsumi Yamaoka wrote:

> (setq w3m-goto-article-function #'browse-url)

Re: Patch to change mouse binding for html view in gnus

[Tom Tromey]

>>>>> "Katsumi" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

Katsumi> For this reason, I recommend using the function form
Katsumi> (lambda (url) (browse-url url) t)
Katsumi> rather than just using `browse-url'.

I tried:

(setq w3m-goto-article-function (lambda (url) (browse-url url) t))

but this fails:

  fboundp((lambda (url) (browse-url url) t))
  w3m-safe-view-this-url()
  w3m-mouse-safe-view-this-url((mouse-2 (#<window 1388 on *Article*> 436 (111 . 209) -81415166 nil 436 (18 . 16) nil (3 . 1) (6 . 13))))
  call-interactively(w3m-mouse-safe-view-this-url)

The reason is this code in w3m-safe-view-this-url:

     (url (or (when (fboundp w3m-goto-article-function)
		(funcall w3m-goto-article-function url))

So to make this work I had to make a new defun and then setq
w3m-goto-article-function to that.  Using an anonymous function won't
work, due to the fboundp.

I didn't look to see whether your patch here fixed this problem or
not, I just thought I'd post in case anybody else is trying to set
things up this way.

>> If you do all your browsing in Emacs then you probably already have
>> browse-url-browser-function set as you like...

Katsumi> I leave all those as the default and have no problem (Firefox
Katsumi> happens to be used by default).  I don't always necessarily use
Katsumi> emacs-w3m.   I use Firefox when purchasing something, manipulating
Katsumi> my bank account... ;-)

Ok.  I suppose this response means you are not swayed by my argument
that users should only have to customize in a single place.

Tom

Re: Patch to change mouse binding for html view in gnus

[Stefan Monnier]

>   fboundp((lambda (url) (browse-url url) t))

The code should use `functionp' instead of `fboundp'.


        Stefan

Re: Patch to change mouse binding for html view in gnus

[Katsumi Yamaoka]

>>>>> Stefan Monnier wrote:

>>   fboundp((lambda (url) (browse-url url) t))

> The code should use `functionp' instead of `fboundp'.

Yup.  I've fixed it about a week ago in the emacs-w3m CVS trunk[1].
Furthermore, I also made a change so that just `browse-url' rather
than such a lambda form for `w3m-goto-article-function' may work[2].

[1] See http://emacs-w3m.namazu.org/index-en.html, or simply get:
    http://cvs.namazu.org/emacs-w3m.tar.gz

[2] http://article.gmane.org/gmane.emacs.devel/74274

>>>>> Tom Tromey wrote:

[...]

Katsumi> I leave all those as the default and have no problem (Firefox
Katsumi> happens to be used by default).  I don't always necessarily use
Katsumi> emacs-w3m.   I use Firefox when purchasing something, manipulating
Katsumi> my bank account... ;-)

> Ok.  I suppose this response means you are not swayed by my argument
> that users should only have to customize in a single place.

That is that at least I am only not motivated to do that.
Neither this represents the emacs-w3m team nor I am a chief of
the team.  You can contact the team at a suitable place, i.e.,
the emacs-w3m mailing list for example.

Regards,

Re: Patch to change mouse binding for html view in gnus

[Graham Murray]

Tom Tromey <tromey@redhat.com> writes:

> Maybe there are people who want to usually use an external browser but
> then browse using w3m from gnus.  That seems weird to me :), but if
> such people exist then, yeah, this should be customizable.

The time that can be useful, if I understand correctly, is when you use
an external browser when actually sat at the machine running gnus, but
also connect into the system using ssh from remote locations and run
gnus in a terminal session. I suspect that there is no way that gnus
could be requested to open a browser local to the system from which you
are ssh'ing, so using w3m on the terminal interface is the only option.

Re: Patch to change mouse binding for html view in gnus

[Tom Tromey]

>>>>> "Graham" == Graham Murray <graham@gmurray.org.uk> writes:

>> Tom Tromey <tromey@redhat.com> writes:
>> Maybe there are people who want to usually use an external browser but
>> then browse using w3m from gnus.  That seems weird to me :), but if
>> such people exist then, yeah, this should be customizable.

Graham> The time that can be useful, if I understand correctly, is
Graham> when you use an external browser when actually sat at the
Graham> machine running gnus, but also connect into the system using
Graham> ssh from remote locations and run gnus in a terminal
Graham> session. I suspect that there is no way that gnus could be
Graham> requested to open a browser local to the system from which you
Graham> are ssh'ing, so using w3m on the terminal interface is the
Graham> only option.

Yes, I agree.  However, this is a generic problem with buttonized URLs
in Emacs, not something specific to the case of using w3m to render
HTML email in Gnus.  In my view, if this problem is to be fixed,
ideally it ought to be fixed in a generic way -- that is, in
browse-url.

Tom