Re: [Usability]Window titles and WM_* properties

[Gregory Merchan <merchan phys lsu edu>]

On Fri, Mar 01, 2002 at 03:52:35PM -0500, Havoc Pennington wrote:
> 
> Gregory Merchan <merchan baton phys lsu edu> writes:
> > > 4.1.2.1.  WM_NAME Property
> >
> 
> Sorry, by NAME I meant the name portion of WM_CLASS.
> 
> > > This property must be present when the window leaves the
> > > Withdrawn state and may be changed only while the window is
> > > in the Withdrawn state.  
> 
> This is the part I mean about how you can't change it at runtime.

That's part of the WM_CLASS. It's basically the name of the program and
there's no reason to change it . . . pretty much ever.


> > As for the WM_WINDOW_ROLE, I never said anything about making it visible.
> > Nonetheless:
> 
> Nonetheless what? What info there did you feel was relevant?

Nonetheless I quote the section because hardly anyone bothers to read 
the ICCCM in the first place. The relevant information is the description
of the property so that anyone who might try to use it can see what it's for.

> Making up nonstandard window titles based on non-internationalized
> properties that are already used for other purposes is just a bad
> idea. Post the idea to wm-spec-list if you want WM authors other than
> me to tell you the same thing.

UTF8_STRING is a valid property type for WM_NAME. The name of an application
is a proper name and not something to translate, so that WM_CLASS must
be a STRING type will rarely matter. As I stated in the original email,
that it must be STRING is unfortunate, but extensions were beyond the scope
of that mail.

As for WM authors: AFAIK, there may be only 2 that have written ICCCM 2.0
compliant window managers. Those are the authors of FVWM and MWM.

Metacity, Sawfish, E16, PWM, TWM, KWM, and surely others I've forgotten
were checked a few weeks ago and all failed a simple test because they don't
manage WM_Sn.

> It would be nice to have a mechanism for constructing titles in a
> standard fashion 

You do. I've shown you.

>            . . . - e.g. if the window manager could do 
> "sprintf ("%s - %s", appname, documentname)" or something - 

That's a lousy way to name a window, even if Netscape 4.x does so.
The document name should come first.

> but it requires new window manager hints for "appname" and
> "documentname", not bogus interpretations of old hints.
> 
> Havoc

It does not require that. They are not bogus.


Greg Merchan