WM: grouping applications (TAI)

[Matthew Woehlke]

Maciej Pilichowski wrote:
> TAI/GAI reactivation ;-)

Yay! :-D

> So I ask you this:
> a) let's say you already have organized TAI, something similar to 
> Konq. TDI but each tab is distinct app, no other app is running -- is 
> task switching is possible via the same alt+tab as usual? Or should 
> switching within tab should be distinct from "normal" task switching?

Well... ideally I would say we should co-opt tab-switching shortcuts. 
The problem is, there don't seem to be any /consistent/ such shortcuts 
(except, annoyingly enough, ctrl-shift-arrows to rearrange tabs), so we 
probably need to define those at some point. Although... TBH it makes 
more sense to use alt-tab, just that users won't expect that.

In short, it's a good question that needs an answer. Perhaps best is to 
us alt-arrows for tree navigation, and alt-tab to switch only top-level 
windows/containers (and I suppose each container remembers its active 
child, same as active widget is remembered). Left/right would be 
configurable to either switch instantly between siblings i.e. like 
current tab switching, or pop up an alt-tab-like tree switcher (or I 
suppose, whatever people come up with for fancy switchers). Down/up 
would go immediately to the switcher, with up starting at the active 
window's parent.

> b) can GAI container can be put (I ignore for the moment _how_) into 
> TAI? 
> 
> c) can TAI container can be put into GAI?

Yes and yes. I started a reply to elsewhere in the thread(s) but might 
as well post my reiterated opinion on nesting here...

 From a technical standpoint, a group (which can be floating windows, 
recall) should be generic. That is, anywhere you need a container 
(including the root), it can be /any/ container. That means that nesting 
is required, since the default configuration needs to be for the root 
container to be floating-windows if we're to have any generic uptake, 
and of course you need at least one level of nesting or this is a waste 
of time ;-).

 From a technical standpoint, my answer is a very emphatic "yes". From a 
usage standpoint, I'm more inclined to take a wait-and-see attitude; we 
can, if it seems needed, impose a depth limit (of one, basically) on 
nesting... although I'll mention again that I don't like artificial 
limits :-).

> PS. All I can say for the time that passed, that I miss GAI for 
> sure ;-)

Yeah. Heck, just a tiling WM would be nice... but of course I don't want 
*only* a tiling WM (and the grouping part of GAI, i.e. groups are 
raised/lowered as one, would be nice also).

-- 
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
-- 
ENOWIT: .sig file not available