Re: [Usability] Grouping Windows: Sticky Windows

[Ryan McDougall <ryan mcdougall telusplanet net>]

On Tue, 2004-04-06 at 10:50 -0700, Nadyne Mielke wrote:
> At 09:03 PM 4/5/2004, Ryan McDougall wrote:
> 
> [snip]
> 
> >While one might be annoyed by windows that are stuck together against
> >your will, it doesn't get too far in the way of usability.
> 
> I'm not sure about that.  Let's assume that I have one main task window 
[snip]
> If I move my main task 
> window towards the various other open windows, they'd all stick together.

"Accidental stickings" are definitely a bad thing, and any design should
avoid them; I'm just saying they don't directly impede usability (ie: it
doesn't stop you from using your email).

> 
> There's the sheer annoyance of having to unstick several windows because I 
> moved one big window.   But worse, I'll have lost my organization of the 
> peripheral windows.  I tend to keep each of those windows in a specific 
> location.  I would lose a lot of time either unsticking them (unless there 
> was an 'unstick all' optin?) and moving them back to where I usually keep 
> them, or in trying to figure out where they are in the new layout.

If "Stickiness" is a modal feature, you could never inadvertently stick
things without having just asked the WM to do so. Example: I want to
stick my terminal and editor, so I click on the terminal window, choose
"stickify this window". The window changes visual (border becomes
greenish?) and when we drag it, it attaches itself to the nearest window
but doesn't actually stick until we "drop" it where we want it (on the
editor). Once dropped the terminal is "stuck" to the editor, but loses
its "stickiness" and will no longer attach to another window.

> 
> I suppose that one way around that problem would be to have a way to 
> specify in a particular window that I never want to have other windows 
> stick to it.  A more general case would be to say that I don't want any 
> more windows to stick to this particular grouping (where the group is n 
> windows, n>=1).  That allows me to say that I want only my network monitor 
> and system monitor to stick together, but if I move one of my IM windows 
> down near them, it won't stick there.  This suggestion might add a lot of 
> conceptual overhead, though.

This is a non-issue with the above method, since windows are not
automatically sticky, only certain user-specified windows temporarily
have this mode.

> 
> [snip]
> 
> >-GUI desktops generate a lot of windows, and manipulating each as
> >separate entities divorced from any relationship with any other window
> >requires the constant mental overhead of know which window relates to
> >which, and the constant manual manipulation (moving, resizing, etc.) on
> >their own. The GIMP is one example, as is the common developer's setup:
> >editor, terminal, debugger, music player, etc. If we have *some* means
> >of tying these windows together we can improve usability by decreasing
> >the distinct and separate entities that a user has to track and
> >manipulate.
> 
> The overridding issue here is that the user has to manually manipulate 
> windows (sometimes repeatedly in the course of a single task) to achieve 
> the highest degree of usability at a system-wide level.  Users have limited 
> screen real estate and a lot of windows vying for both screen placement and 
> user attention.  Further, optional placement on the screen in terms of 
> screen real estate used (that is, few overlapping windows and few unused 
> bits of the desktop) is not necessarily optimal placement on the screen in 
> terms of user attention.
> 
> Assuming that I've stated the issue correctly, I'm not entirely certain 
> that sticky windows is the answer.  It might be a part of the answer, 
> though.  I do like the idea of being able to stick some of my windows 
> together.  But I'm not certain of doing it automatically at a system-wide 
> level.
> 
> Just to be clear, I'm not saying that I don't like the idea.  I'm just 
> trying to get the idea fleshed out, and trying to figure out if there are 
> usability problems that could occur.

I agree.

> 
> /nm 

Cheers,
Ryan