Re: move/resize handling again

[Dominik Vogt <dominik vogt gmx de>]

On Tue, Nov 12, 2002 at 02:35:49PM -0500, Havoc Pennington wrote:
> On Tue, Nov 12, 2002 at 08:16:06PM +0100, Marko Macek wrote:
> Dominik Vogt wrote:
>
> > How unfortunate that the new version just tries to
> > enforce the imagined "right way" of handling the win_gravity in
> > configure requests.  I had at least hoped that it is somewhere
> > mentions that most windows managers and clients traditionally do
> > not follow this "rule".  Sigh, I guess I have to continue to
> > ignore much of the EWMH spec in order to write a felxible window
> > manager.
> > 
> > Sorry, I just don't see how this is going to work.
> > 
> > Either the EWMH and/or the ICCCM conflict and one of them (EWMH?) is 
> > broken, or your window manager is.
> >
> > A "legacy" mode (ignoring parts of ICCCM and/or EWMH) is unfortunatelly 
> > not possible since there is no way to identify applications that are not 
> > ICCCM/EWMH compliant.

Of course we can not change every legacy application.  But the spec
can state:

 - An EWMH compliant application MUST conform to the ICCCM way.
 - A compliant application MUST set some property of its window to
   document compliance.
 - A compliant window manager MUST conform to the ICCCM way when
   it detects said property on a client window.
 - If the property is not present on a window the window manager
   SHOULD use the legacy way as there are many applications that
   do not comply to the ICCCM rule.
 - Applications that are currently not EWMH compliant SHOULD be
   changed to become ICCCM compliant and set said property, or do
   not set it and use the legacy way (not recommended).

Okay, it's silly for applications to say "hey, I'm *not* broken",
but the WM needs a hint here and I don't see the legacy apps
providing it.

> Exactly, whenever you have a protocol and various people have
> interpreted it in different ways, and you don't have a way to know how
> the other end has interpreted it, the only solution is to agree on one
> interpretation and then start fixing code. If there were a way to
> magically do the right thing, then we would not have a problem, and
> would not have any need to clarify the interpretation.

Yes, of course.  Normally, I wouldn't talk about this problem so
much.  But it annoys so many users and there are so many legacy
applications out there

> We have icewm, sawfish, metacity, kwin at least all working the same
> way now, right? Maybe also blackbox? That's pretty good.

But it doesn't solve the problem.  It's not about which window
manager supports which behaviour.  It's the clients that have to
be fixed.  Just saying "this way is right and the other isn't"
will just not help to write a window manager that is happy with
all but the most broken apps.

> GTK also
> follows the EWMH interpretation on the client side, and I would expect
> that Qt does as well.

And probably it is easy in Qt and Gtk to add that property on a
client window.

Bye

Dominik ^_^  ^_^

 --
Dominik Vogt, dominik vogt gmx de
Reply-To: dominik vogt gmx de