Comments on 1.9c

[jg pa dec com (Jim Gettys)]

Note: I never paid too much attention to ICCCM, so my background
is questionable on the details: it would be good to get Dave Rosenthal
to do a detailed read...  Dave, are you listening?

I find the spec not very rigorous: in particular, it isn't clear
who the players are (wm, client application, panel, etc.),
and who does what when to whom.

I recommend use of the terminology used in RFC 2119, along with making
it clear what the responsibility of each class of client is in the
manipulation of each property. See: ftp://ftp.isi.edu/in-notes/rfc2119.txt

Aren't some of the "Implementation Notes" at the end really more of an 
introduction to what the protocol is attempting to achieve?  Shouldn't
those become an introduction?

Packaging the hints into 2 X properties for performance.
--------------------------------------------------------

Don't do this!!!! There is a fundamental conceptual mistake lurking here: 
you are equating a property with a round trip. Properties should contain 
related information, with provision for extension on each property whenever 
possible. Glomming together unrelated data into a single property will 
make things very difficult to evolve. 

If round trips are a problem to fetch multiple properties, you
write yourself an Xlib routine to retrieve multiple properties
in one round trip.

This isn't rocket science.  There are lots of examples of this in Xlib.
There are a few things that make this more painful than it might be,
but it isn't all that hard...  Don't make a mess of things to avoid
straightforward code...

_NET_WM_ICON

What about "Icon windows"?  It should be possible for apps to provide
windows for icons, and therefore provide animated icons.  Prior to X11,
this was possible, and it turned out to be very useful: xterm had a 1x2
high pixel font, and you could see when the xterm changed while running.
I've always missed this feature.

The definition of  ARGB image type for icons isn't very clear.  Question:
is there some existing image format that can be adopted "whole hog" rather
than introducing another one?


			 - Jim