Re: _NET_WM_MOVERESIZE

[Sasha_Vasko osca state mo us]

> Hi,
>
> Shouldn't the message contain a) a timestamp when the button press
> occurred, so the grab can be done with that timestamp and b) the
> button which was pressed, so the WM knows which release signals the
> end of the drag?

Yes the timestamp is mandatory. As for button - it better be
buttons+modifier
mask ( as returned by XQueryPointer ), since it is concievable that
something like Meta+Button1 will be used for move/resize.
Window Manager will then need to establish active grab using XGrabPointer
and then XQueryPointer to confirm that specifyed combination is still
pressed.

>
> Also, I know Owen and Matthias were talking a bit about X properties
> containing information such as double-click time, to be shared among
> toolkits. The window manager should probably also share this info
> (e.g. doubleclick time), and probably the button to be used for
> move/resize should be one of those settings.

It is a good idea to have such property for dubleclick time, since it is
very simple, yet crucial thing and it better be consistent throghout the
server.

As for the button for move/resize - it is different - there could be
several
button+modifier combos assigned for any particular task, so storing it in
the
property will make it kinda cumbersome. If one really wants to have this
sort of
things stored in properties - it better be comprehensive feel database -
like
list of all the pointer+keyboard combos handled by Window manager on root
and
client windows, possibly with some information in regards what those things
are
used for. I can actually think of a few scenarios where it might be
usefull.

> Havoc

Regards
Sasha Vasko