GTK3 Emacs

GTK3 Emacs

[joakim]

I noticed that is now possible to compile Emacs with GTK3. It looks
pretty good! Thanks guys!
-- 
Joakim Verona

Re: GTK3 Emacs

[Antoine Levitt]

30/06/11 22:43, joakim@verona.se
> I noticed that is now possible to compile Emacs with GTK3. It looks
> pretty good! Thanks guys!

Does it actually change anything?

Re: GTK3 Emacs

[joakim]

Antoine Levitt <antoine.levitt@gmail.com> writes:

> 30/06/11 22:43, joakim@verona.se
>> I noticed that is now possible to compile Emacs with GTK3. It looks
>> pretty good! Thanks guys!
>
> Does it actually change anything?


Just the looks AFAICT. However, GTK3 has many differences at the API
level. I noticed this by trying to convert my xwidget branch to GTK3 and

I had to make many changes in my code, so the overall effort to convert all of
Emacs to GTK3 must have been big. 

As an aside, some of the changes in GTK3 are annoying. The method of
clipping I employ in GTK2 no longer works for innstance.
-- 
Joakim Verona

Re: GTK3 Emacs

[Jan Djärv]

2011-07-01 00:52, Antoine Levitt skrev:
> 30/06/11 22:43, joakim@verona.se
>> I noticed that is now possible to compile Emacs with GTK3. It looks
>> pretty good! Thanks guys!
>
> Does it actually change anything?
>

Not much.  But AFAIK, Gtk3 themes are not compatible with Gtk2 themes, so 
Emacs compiled for Gtk2 won't use the right theme on systems with Gtk3 (and 
vice versa I guess).

	Jan D.

Re: GTK3 Emacs

[Jan Djärv]

2011-07-01 08:42, joakim@verona.se skrev:
> Antoine Levitt<antoine.levitt@gmail.com>  writes:
>
>> 30/06/11 22:43, joakim@verona.se
>>> I noticed that is now possible to compile Emacs with GTK3. It looks
>>> pretty good! Thanks guys!
>>
>> Does it actually change anything?
>
>
> Just the looks AFAICT. However, GTK3 has many differences at the API
> level. I noticed this by trying to convert my xwidget branch to GTK3 and
>
> I had to make many changes in my code, so the overall effort to convert all of
> Emacs to GTK3 must have been big.
>
> As an aside, some of the changes in GTK3 are annoying. The method of
> clipping I employ in GTK2 no longer works for innstance.

As Gtk3 now is all Cairo, I think you must do clipping on that level.
Or is that behaving differently in Gtk2?
It might also be that more widgets are windowless (i.e. writes into the parent 
window).

For Emacs, scroll bars may not be redrawn correctly for Gtk3, so there is 
definitly some changes in drawing.

	Jan D.

Re: GTK3 Emacs

[joakim]

Jan Djärv <jan.h.d@swipnet.se> writes:

> 2011-07-01 08:42, joakim@verona.se skrev:
>> Antoine Levitt<antoine.levitt@gmail.com>  writes:
>>
>>> 30/06/11 22:43, joakim@verona.se
>>>> I noticed that is now possible to compile Emacs with GTK3. It looks
>>>> pretty good! Thanks guys!
>>>
>>> Does it actually change anything?
>>
>>
>> Just the looks AFAICT. However, GTK3 has many differences at the API
>> level. I noticed this by trying to convert my xwidget branch to GTK3 and
>>
>> I had to make many changes in my code, so the overall effort to convert all of
>> Emacs to GTK3 must have been big.
>>
>> As an aside, some of the changes in GTK3 are annoying. The method of
>> clipping I employ in GTK2 no longer works for innstance.
>
> As Gtk3 now is all Cairo, I think you must do clipping on that level.
> Or is that behaving differently in Gtk2?
> It might also be that more widgets are windowless (i.e. writes into
> the parent window).

Yes, I written some in the xwidget readme about it. The issue is mith
windowless widgets. They seem to simply ignore clipping area
requests. In GTK2 I was able to use GtkFixed and configure it to be
windowed. This no longer works in gtk3. 

>
> For Emacs, scroll bars may not be redrawn correctly for Gtk3, so there
> is definitly some changes in drawing.



>
> 	Jan D.

-- 
Joakim Verona

Re: GTK3 Emacs

[Jan Djärv]

2011-07-01 10:13, joakim@verona.se skrev:
> Jan Djärv<jan.h.d@swipnet.se>  writes:
>
>> 2011-07-01 08:42, joakim@verona.se skrev:
>>> Antoine Levitt<antoine.levitt@gmail.com>   writes:
>>>
>>>> 30/06/11 22:43, joakim@verona.se
>>>>> I noticed that is now possible to compile Emacs with GTK3. It looks
>>>>> pretty good! Thanks guys!
>>>>
>>>> Does it actually change anything?
>>>
>>>
>>> Just the looks AFAICT. However, GTK3 has many differences at the API
>>> level. I noticed this by trying to convert my xwidget branch to GTK3 and
>>>
>>> I had to make many changes in my code, so the overall effort to convert all of
>>> Emacs to GTK3 must have been big.
>>>
>>> As an aside, some of the changes in GTK3 are annoying. The method of
>>> clipping I employ in GTK2 no longer works for innstance.
>>
>> As Gtk3 now is all Cairo, I think you must do clipping on that level.
>> Or is that behaving differently in Gtk2?
>> It might also be that more widgets are windowless (i.e. writes into
>> the parent window).
>
> Yes, I written some in the xwidget readme about it. The issue is mith
> windowless widgets. They seem to simply ignore clipping area
> requests. In GTK2 I was able to use GtkFixed and configure it to be
> windowed. This no longer works in gtk3.
>

Have you tried to use GtkEventBox?

	Jan D.

Re: GTK3 Emacs

[Ken Raeburn]

On Jun 30, 2011, at 16:43, joakim@verona.se wrote:
> I noticed that is now possible to compile Emacs with GTK3. It looks
> pretty good! Thanks guys!

By any chance does GTK3 fix the multiple-display bugs that keep some of us using the non-GTK configuration?  (From previous discussions, I expect this is a long shot...)

Ken

Re: GTK3 Emacs

[joakim]

Ken Raeburn <raeburn@raeburn.org> writes:

> On Jun 30, 2011, at 16:43, joakim@verona.se wrote:
>> I noticed that is now possible to compile Emacs with GTK3. It looks
>> pretty good! Thanks guys!
>
> By any chance does GTK3 fix the multiple-display bugs that keep some of us using the non-GTK configuration?  (From previous discussions, I expect this is a long shot...)

No idea. I can test it if you tell me the bug number!


>
> Ken

-- 
Joakim Verona

Re: GTK3 Emacs

[David De La Harpe Golden]

On 05/07/11 07:23, joakim@verona.se wrote:
> Ken Raeburn<raeburn@raeburn.org>  writes:
>
>> On Jun 30, 2011, at 16:43, joakim@verona.se wrote:
>>> I noticed that is now possible to compile Emacs with GTK3. It looks
>>> pretty good! Thanks guys!
>>
>> By any chance does GTK3 fix the multiple-display bugs that keep some of us using the non-GTK configuration?  (From previous discussions, I expect this is a long shot...)
>
> No idea. I can test it if you tell me the bug number!
>
Presumably the one gtk2 emacs daemon mode now warns about.

-8<----------

david@golden1:~$ emacs --daemon

Warning: due to a long standing Gtk+ bug
http://bugzilla.gnome.org/show_bug.cgi?id=85715
Emacs might crash when run in daemon mode and the X11 connection is 
unexpectedly lost.
Using an Emacs configured with --with-x-toolkit=lucid does not have this 
problem.

-8<----------

If the problem is still present in gtk3, and a comment on the bug 
suggests it is - https://bugzilla.gnome.org/show_bug.cgi?id=85715#c55 , 
I guess gtk3 emacs daemon mode should spit out a similar warning (if it 
doesn't already, I haven't got as far as trying a gtk3 emacs build yet).

Re: GTK3 Emacs

[joakim]

David De La Harpe Golden <david@harpegolden.net> writes:

> On 05/07/11 07:23, joakim@verona.se wrote:
>> Ken Raeburn<raeburn@raeburn.org>  writes:
>>
>>> On Jun 30, 2011, at 16:43, joakim@verona.se wrote:
>>>> I noticed that is now possible to compile Emacs with GTK3. It looks
>>>> pretty good! Thanks guys!
>>>
>>> By any chance does GTK3 fix the multiple-display bugs that keep some of us using the non-GTK configuration?  (From previous discussions, I expect this is a long shot...)
>>
>> No idea. I can test it if you tell me the bug number!
>>
> Presumably the one gtk2 emacs daemon mode now warns about.
>
> -8<----------
>
> david@golden1:~$ emacs --daemon
>
> Warning: due to a long standing Gtk+ bug
> http://bugzilla.gnome.org/show_bug.cgi?id=85715Emacs might crash when run in daemon mode and the X11 connection is
> unexpectedly lost.
> Using an Emacs configured with --with-x-toolkit=lucid does not have
> this problem.
>
> -8<----------
>
> If the problem is still present in gtk3, and a comment on the bug
> suggests it is - https://bugzilla.gnome.org/show_bug.cgi?id=85715#c55, I guess gtk3 emacs daemon mode should spit out a similar warning (if
> it doesn't already, I haven't got as far as trying a gtk3 emacs build
> yet).

The problem perists sadly. I did get the same warning also.

This was sad, I used mtty-emacs a lot before, but not so much lately, so
I suppose thats why I didnt notice it.

I suppose you've alredy looked at using NX or VNC as a workaround.

-- 
Joakim Verona

Re: GTK3 Emacs

[Ken Raeburn]

On Jul 5, 2011, at 02:23, joakim@verona.se wrote:
> Ken Raeburn <raeburn@raeburn.org> writes:
>> By any chance does GTK3 fix the multiple-display bugs that keep some of us using the non-GTK configuration?  (From previous discussions, I expect this is a long shot...)
> 
> No idea. I can test it if you tell me the bug number!

It's Emacs bug numbers 1493, 3360, and others.  Also https://bugzilla.gnome.org/show_bug.cgi?id=85715 which has a comment from 3 days ago saying the bug is still present in GTK+ 3.0.11.  :-(

Ken

Re: GTK3 Emacs

[Chong Yidong]

joakim@verona.se writes:

> The problem perists sadly. I did get the same warning also.

That's because the warning comes from a workaround in the Emacs sources,
which kills Emacs rather than allowing it to loop forever.

To test this properly, you need to remove that workaround and see if
Emacs now recovers gracefully, or whether it still loops forever.

Re: GTK3 Emacs

[Ulrich Mueller]

>>>>> On Tue, 05 Jul 2011, Chong Yidong wrote:

>> The problem perists sadly. I did get the same warning also.

> That's because the warning comes from a workaround in the Emacs
> sources, which kills Emacs rather than allowing it to loop forever.

In xterm.c, function x_connection_closed(), if I got it right.

> To test this properly, you need to remove that workaround and see if
> Emacs now recovers gracefully, or whether it still loops forever.

When the X connection is lost, e.g. by logging out the user, Emacs
will loop, taking all available CPU time.

Tested with revision 104968 (BZR trunk as of today) and GTK+ 3.0.11 on
an x86_64 GNU/Linux system.

Ulrich