[IAEP] Running regular X11 apps

[Sayamindu Dasgupta]

On Fri, Jun 27, 2008 at 9:57 PM, Marco Pesenti Gritti
<mpgritti at gmail.com> wrote:
> On Fri, Jun 27, 2008 at 6:13 PM, Sayamindu Dasgupta <sayamindu at gmail.com> wrote:
>> On Fri, Jun 27, 2008 at 9:40 PM, Marco Pesenti Gritti
>> <mpgritti at gmail.com> wrote:
>>> On Fri, Jun 27, 2008 at 6:01 PM, Jim Gettys <jg at laptop.org> wrote:
>>>> On Fri, 2008-06-27 at 11:19 +0200, Marco Pesenti Gritti wrote:
>>>>
>>>>> There are ways we can implement different behavior for different
>>>>> windows (for example the maximus approach could be extended to do so).
>>>>> The problem is, how do we know which windows should be displayed
>>>>> fullscreen (say the firefox or the gedit one) and which not (the
>>>>> gimp)? afaik there is no window hint which could help us there...
>>>>>
>>>>
>>>> There are ICCCM/EWMH hints to request full screen behavior, already
>>>> implemented by window managers: e.g. totem or other video players use
>>>> these routinely.
>>>
>>> Here we was discussing the approach used by maximus. They are
>>> basically maximizing/undecorating windows whenever they are mapped (as
>>> long as they are normal windows and not dialogs). Now this works for
>>> something like firefox, but for gimp there is no hint that tells you
>>> that the tools window should *not* be maximized.
>>>
>>> About the fullscreen hint. Sayamindu tried out that approach but he
>>> run into problems. One of them was that panels are hidden by
>>> fullscreen windows (and we made the sugar frame a panel). But I think
>>> there was others.
>>>
>>
>> Fullscreen seems to be significantly different from maximized +
>> undecorated. The only sugar activity I can think of which uses
>> fullscreen mode at certain times is Record.
>
> All the activities can use fullscreen actually, it's in the base
> class. In matchbox the effect is just that the toolbars goes away, but
> if you runned them on a standard desktop they would actually be made
> fullscreen.
>
> My feeling is that the right approach here is to let the window
> manager deal with the window sizing, on the base of the available
> hints. That ways both activities and applications can behave in an
> optimal way, depending on the window manager they run in.
>
> The simplest implementation of this approach is maximus, i.e. the
> window manager maximize all normal windows. This should work well for
> most existing applications. For the side cases you'd have either to
> get additional hints in the freedesktop spec or to use heuristics like
> those David described.
>
> Does it make sense? Can you think of better approaches?
>

It does make sense. I think metacity already does special stuff for
certain applications (eg: gnome-terminal[1]). From what I understand,
the number of apps which would require "special treatment" should not
be very high.

Cheers,
Sayamindu



[1] http://git-mirror.gnome.org/?p=metacity;a=blob;f=src/core/window.c#l1960

-- 
Sayamindu Dasgupta
[http://sayamindu.randomink.org/ramblings]