Re: _NET: Disabling shading

["Denis O. Mikhalkin" <dom sparc spb su>]

On Wed, 2003-10-01 at 18:40, Dominik Vogt wrote:
> On Wed, Oct 01, 2003 at 06:23:38PM +0400, Denis O. Mikhalkin wrote:
> > On Wed, 2003-10-01 at 18:16, Matthias Clasen wrote:
> > > > > 
> > > > > This is the big misunderstanding. It's the users desktop and they should
> > > > be
> > > > > in control of the windows appearing on it, with the help of a wm which
> > > > they've
> > > > > selected for its interpretation of the specs.  
> > > > I don't agree with that. I don't think user want to do what you say -
> > > > users usually want PREDICTABILITY.
> > > 
> > > What you propose only adds predictability for the developer how exactly 
> > > "his" windows will behave.  Which might in fact be of some value if he needs
> > > 
> > > to specify how exactly to do window management tasks on "his" windows.
> > > But why the hell should he, instead of just referring to the wm manual ? 
> > Huh, wm manual? Is there any? But there are numerous software manuals,
> > and numerous software products, and if some window of the program
> > behaves specially the author can mention this in the manual since anyway
> > they describe how to operate with different windows.
> > 
> > > It is far better to ensure desktop-wide uniformity and learnability by
> > > letting the wm set the policy how windows of certain types are managed.
> > This uniformity is achieved by restriction which people don't accept.
> > Instead of using WMs they: 
> > - switch to another WM, since it provides the features they wanted
> > - write hacks
> > Is that what you want? I don't think it is hard to even go half-way - to
> > provide features and to have one value called "WM-specific".
> 
> > But I doubt anyone will use it
> 
> If there is no interest in different policies, then how do you
> explain the existence of more than 100 X window managers?
> 
> > developers want predictability.
> 
> And yet they refuse to adhere to the standards that are already
> there. 
I don't see this as an argument - if you look at the source code of any
freely available WM then you'll see that they violate a lot of
standards. As I said - if you restrict, you get what you want - people
hack. That's what happens when stadnards are not flexible.

> You should note that (especially commercial) applications
> violate or ignore the rules far more often than window managers.
I don't see this. Commercial application don't do this for fun - but
because of the high needs, keeping in mind the consequences. Most
commercial applications are specifically declared "compliant" to some
protocols so they follow protocols more often. While most people do this
becuase it looks or behaves the way it pleases them - and do this often.

> And, by the way, it is foolish of a deveoper to think she knows
> best how the desktop must behave.  This "shading" discussion is
> the perfect example.  Users love predictability, and they
> absolutely *hate* if every window behaves in a different way. 
Again, even if you restrict the developer to only allowing him to create
windows then either your WM won't survive or people will find hacks.
Instead, you should enable features, control correctness, and describe
how good behaving should look like. People are not stupid - when they
see  where the flow goes they usually follow.

>  The fvwm mailing list is already full of mails a la "why can't I
> iconify/resize/maximise/... this specific window?" 
This is because the author of whatever toolkit that is didn't declare
that the window can't be "iconify/resize/maximise/". This is exactly we
are against of - everything should be specified. And we receive numerous
complaints that our windows ARE being
focused/minimized/resizable/decorated or ARE NOT some of these while we
declared that they will be and it happens because some hints/state
hasn't been specified correctly so some freedom of interpretation has
been left either for us or WM. We are against this - freedom here should
be in the wide variaty of possibilities, not in a wide range of
interpretations.

>  So what would a "disallow shading" hint add to predictability? 
It will add that our windows won't be shaded(I am speaking about Java).
So we can predict that, and we can tell developers "your window won't be
shaded". And they write in their documentation "this window won't be
shaded". 

>  What's important for an application may be completely irrelevant for the app's
> user. The developer can not decide that.
What do you say? Do I not know how my office application should work? Do
I not know which windows it shows and how? Definitely, WM can't know
this, while I do, and user will after he works with my application for
some time or reads documentation. What a nice surprise it will be when
after that the program will start to work differently because WM decides
to display windows differently.

Denis