Re: Spec Purpose and Definition

[Michael ROGERS <M Rogers cs ucl ac uk>]

>> Set a hint, yes. Maximise its own window, no. That breaks the "chain of
>> command" (user->window manager->application) and allows an application to
>> define its own window layout policy.
>
>It's still a hint, only given while beeing mapped. The wm may obey or ignore
>it. 

The maximise hints are a good idea. I'm talking about the messages sent by
the pager (or other applications) telling the window manager which windows
to raise or maximise, and which desktop to switch to. The window manager has
to obey these, otherwise window manager/pager interaction will be broken.
That's no problem when the pager uses them, but it is a problem if other
apps start using them, because it breaks the window manager's layout policy.

>Applications _should_ be allowed to define its own window layout policy.

No. That is the job of the window manager.

>ICCCM encourages that with the USPos hints, unfortunately ICCCM doesn't have a
>concept of one of the most important attributes: maximized. This is why we 
>have to add it.

IIRC, USPos hints should be used to communicate the user's preferences, not
used by the application to communicate the programmer's preferences. I think
the pager hints should be used in the same way (as the pager uses them, not
as you have suggested knotes would use them).

>> I think if my computer started switching desktops on its own while I was
>> working, whenever an application thought it had something important to tell
>> me, I would get very irritated. However, I know that E provides this option
>> for popup windows. Can't imagine it being anything but annoying...
>
>As I pointed out it's very useful for particular applications, for example
>knotes. Where's you UNIX spirit? If you don't like a particular application,
>don't use it. We are not here to define what is good or evil or what users
>should do and what not. 

OK, good point. If applications misbehave, people won't use them. However, I
still think we should find a better way of getting the user's attention than
automatically switching desktops, or raising a dialog box which gets the
keyboard focus when I'm in the middle of typing. This could be a useful
addition to the spec.

>> I'm not really worried about the three lines of code.  :)  I'm worried about 
>> the idea of an application deciding to override the window manager and choose 
>> its own stacking position, desktop, etc, and changing the active window or 
>> active desktop whenever it likes. This is just going to annoy the user.
>
>Please, let this desicion be made by the application developers. If the user
>feels like this all is crap, he/she can tell the window manager to ignore all
>these hints.

Fair enough.

>Michael, you are basically saying: ICCCM already gives applications too much
>possibilities, why adding even more?

I'm not saying that. There are some important things (like multiple desktop
support) missing from the existing hints. I'm just saying, let's bear in
mind that window managers exist for a reason - so you can have one window
policy which applies to all applications. If we can extend the boundaries of
what applications can do within that policy, great. But let's not add things
which don't make sense.

>Frankly, this is the whole point in discussing on this list. The big bunch 
>of "proprietary" hints shows, that this is more than necessary and so we 
>want to clean up this mess. If you don't agree with this goal, I wonder what 
>your mission here is. We are surely not here to turn back the wheel.

I'm sorry that I've come across as some kind of Luddite. I think it's
important to tidy up the mess of ICCCM and proprietary hints, and I don't
have any vested interest in keeping things the same. Mosquito is not
written yet, so I don't care how much things change! I just think pager
support is messy. Maybe it's not that important, but I didn't see anyone
else representing that view. Anyway, I think it's time I shut up for a while.


Michael Rogers