bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Dan Nicolaescu]

Gtk+-3 will make some structure members private.

Compiling with -DGSEAL_ENABLE points out the problems.

See also http://live.gnome.org/GnomeGoals/UseGseal
for more info

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Jan Djärv]

Dan Nicolaescu skrev 2010-06-24 20.34:
>
> Gtk+-3 will make some structure members private.
>
> Compiling with -DGSEAL_ENABLE points out the problems.
>
> See also http://live.gnome.org/GnomeGoals/UseGseal
> for more info
>
>

It is currently not an issue, configure checks for gtk+-2.0, so 3.0 isn't 
considered yet.  Also, other libraries (rsvg, gconf possibly dbus) must also 
be moved forward.

Apart from that, it is the question of how old Gtk+ we should support. 
Currently we say 2.6 (which is really old) or newer.  Most accessor functions 
in Gtk+ where introduced in 2.14, some in 2.16, and in one case, 2.20.

We could add #ifdef:s, but I rather not do that unless we really need to.

That said, I have a private branch that works with Gtk+ 3.0, and some of the 
fixes there can go in the trunk.  Maybe we should bump required to 2.14 and 
get most of the work checked in.

	Jan D.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Glenn Morris]

Jan Djärv wrote:

> That said, I have a private branch that works with Gtk+ 3.0, and some
> of the fixes there can go in the trunk.  Maybe we should bump required
> to 2.14 and get most of the work checked in.

For the record, RHEL 5, the most recent Red Hat Enterprise Linux, only
has 2.10. I don't have an opinion whether that is important or not.
There will probably be a RHEL 6 (with 2.18) before Emacs 24.1.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Dan Nicolaescu]

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

> Dan Nicolaescu skrev 2010-06-24 20.34:
>>
>> Gtk+-3 will make some structure members private.
>>
>> Compiling with -DGSEAL_ENABLE points out the problems.
>>
>> See also http://live.gnome.org/GnomeGoals/UseGseal
>> for more info
>>
>>
>
> It is currently not an issue, configure checks for gtk+-2.0, so 3.0
> isn't considered yet.  Also, other libraries (rsvg, gconf possibly
> dbus) must also be moved forward.
>
> Apart from that, it is the question of how old Gtk+ we should
> support. Currently we say 2.6 (which is really old) or newer.  Most
> accessor functions in Gtk+ where introduced in 2.14, some in 2.16, and
> in one case, 2.20.
>
> We could add #ifdef:s, but I rather not do that unless we really need to.
>
> That said, I have a private branch that works with Gtk+ 3.0, and some
> of the fixes there can go in the trunk.  Maybe we should bump required
> to 2.14 and get most of the work checked in.

Good to hear you already have some solution.
IMHO the main thing is that emacs should not be seen as lagging
behind.  When the first distribution switches to version 3, it would be
good to have something available...

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Jan Djärv]

25 jun 2010 kl. 23:42 skrev Dan Nicolaescu <dann@gnu.org>:
>>> 
>>> 
>>> 
>>> 
>>> 
>> 
>> 
>> 
>> 
> Good to hear you already have some solution.
> IMHO the main thing is that emacs should not be seen as lagging
> behind.  When the first distribution switches to version 3, it would be
> good to have something available...

I 
 we must see a Gnome based on Gtk 3.0 first. It is not possible (without #ifdefs) to support < 2.14 and 3.0. 

If it is just a few ifdefs I will add them.  But we need to decide if configure should start to check for Gtk 3.0. Maybe now is the time.

     Jan D.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Jan Djärv]

25 jun 2010 kl. 23:02 skrev Glenn Morris <rgm@gnu.org>:

> Jan Djärv wrote:
> 
>> That said, I have a private branch that works with Gtk+ 3.0, and some
>> of the fixes there can go in the trunk.  Maybe we should bump required
>> to 2.14 and get most of the work checked in.
> 
> For the record, RHEL 5, the most recent Red Hat Enterprise Linux, only
> has 2.10. I don't have an opinion whether that is important or not.
> There will probably be a RHEL 6 (with 2.18) before Emacs 24.1.

Good to know. Thanks. 

    Jan D.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Chong Yidong]

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

> 25 jun 2010 kl. 23:42 skrev Dan Nicolaescu <dann@gnu.org>:
>
>> Good to hear you already have some solution.
>> IMHO the main thing is that emacs should not be seen as lagging
>> behind.  When the first distribution switches to version 3, it would be
>> good to have something available...
>
> I we must see a Gnome based on Gtk 3.0 first. It is not possible
> (without #ifdefs) to support < 2.14 and 3.0.
>
> If it is just a few ifdefs I will add them.  But we need to decide if
> configure should start to check for Gtk 3.0. Maybe now is the time.

Yes, I think bumping the requirement to 2.14 in the trunk is fine, as is
adding support for 3.0, provided it does not interfere with the GTK-2
support.  Thanks for working ahead on this.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Yavor Doganov]

Chong Yidong wrote:
> Yes, I think bumping the requirement to 2.14 in the trunk is fine,

That would be unfortunate since gNewSense DeltaH has 2.12.

bug#6505: make the Gtk+ port compile with -DGSEAL_ENABLE in preparation for Gtk+-3

[Jan Djärv]

The required effort to keep supporting 2.6 and newer compared to bumping up to 
2.14 was small (I still had to make backwards compatible macros for functions 
introduced in 2.16, 2.18 and 2.20), so I kept that as minimum version.

Configure now accepts --with-x-toolkit=gtk3 to compile with Gtk+ 3.0 (actually 
2.90 as 3.0 isn't out yet).

	Jan D.