Re: Quick question about spec

On Fri, 18 Feb 2000, Paul Warren wrote:

> On Tue, 15 Feb 2000, Tim Janik wrote:
> > On Sat, 12 Feb 2000, Paul Warren wrote:

> > > I have a todo list of things that I intended to put in, but have not had
> > > time to write up.  Mostly these are suggestions/reminders from Tim Janik:
> > > 
> > 
> > i think that one is still outstanding, what about:
> > 
> > 
> > A property to be set on window manager windows that serve as a new root window,
> > by covering the real root window entirely entirely for a specific viewport or
> > desktop, and by having application windows reparented to them. (This is often
> > used for virtual desktop implementations by window managers.)
> This sounds good, although a brief explanation of the purpose of the
> property would be useful - it is for pager apps to know which windows to
> watch, right?

yes, basically. it is also important for toolkits to figure the position of their
wm frame window, so they can store/restore window positions, since an unmapped window
moved to x,y will be mapped at x+frame_width,y+frame_height.

> > > o Mandate ICCCM compliant window moves.
> > 
> > ok, i do not know where you want to put this, but to try to come up with
> > a good recomendatoin (basically reworded from the relevant mails on that topic):
> > 
> > Window Movement
> > 
> > According to the ICCCM, applications should not see unnecessary differences
> > between running with or without a window manager. Therefore window movements
> > for already mapped windows, such as ones requested by
> > XMoveWindow(Display, Window, X, Y) have to move the window Window to the
> > coordinates (X, Y) and not cause the window's window manager frame window
> > to end up at (X, Y).
> Yep - sounds reasonable.  There is clearly confusion over which of the two
> alternatives is "correct", hence different WMs implement it differently.
> Far more important than getting it right is making it consistent.

well that, or demand non-compliant wms to set a property. but i'd much prefer
simply mandating ICCCM compliance with the wm spec doing some clarification on
that issue.

> > > o Detection of compliant Window Managers - should set a property on WM
> > > owned window, to avoid stale root window property if the WM dies
> > > unexpectedly.
> > 
> > yes, that requires only slight changes to the wm identification property.
> So, is there any problem with reusing the existing method described in
> the old Gnome spec at:
> (apart from changing the property names...)

it'd be good if the window managers would also set a _NET_WM_NAME
(or _NET_WM_ID) property on their
_NET_SUPPORTING_WM_CHECK window, so applications like the control center
could figure *what* window manager is currently running.

> > > o Expand on write up of _NET_ACTIVATE.
> > 
> > what was that for?
> Sorry, that should be _NET_ACTIVE_WINDOW, and more specifically the
> degrees of activation, as described on:
> What is the purpose of the degrees of activation?

i don't know, who braught that issue up?

> > > o Context help - implement basic protocol for WM to invoke context help on
> > > client window.
> > 
> > has anyone suggested an implementation for that at all?
> Yep, Matthias suggested a wm protocol.  See:
> and Owen Taylor then suggested adapting the xdnd protocol to this
> purpose:
> I'm not familiar with the xdnd protocol, but what he suggests sounds
> good - allowing the application to give some feedback as to whether the
> user is going to get any help when they click.  
> We never got any feedback on this idea.  I would like to include both
> possibilities - the Matthias' suggestion is pretty simple to implement
> on the WM.  The best way to do Owen's suggestion involves the WM
> implementing some of the Xdnd protocol, which I suspect some WM authors
> would be reluctant to implement.  It would be nice to have the first
> protocol as a fall back for the second.

i'm not so sure specifying two protocols is a good idea. we've had some lengthy
discussions on this on gnome-hackers and on irc, and basically something like
the Xdnd protocol is required for sane context sensitive help (having no
feedback is really annoying for newbies, and they are actually the ones that
need this feature the most). however, you also have a valid point with wm
authors may be not willing to implement the required overhead, at least not
at this point. so because this issue hasn't yet been finally settled,
my personal feeling is, we should leave it out of the spec for now.
we can still have an addendum at a later point.

> Paul


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]