Re: Pending 1.2 stuff

[Michael Toomim <toomim uclink4 berkeley edu>]

Matthias Clasen wrote:
> I must admit that I don't understand the issue at all. I thought that
> we indroduce the inter-desktop geometry to get aways from the
> 1-dimensional navigation like "previous desk" or "next desk". So, as
> long as the layout is 2x4, a user shouldn't have to care if the desk
> to the left of the current one is desk 5 or desk 2 or desk 17. 

Good point!  If a user is using a 2-d virtual desktop, the actual 
numerical ID's for workspaces shouldn't matter, and shouldn't even have 
to be user-visible information.  In particular, a workspace ordering 
that reads like a book (ie. increasing by left-to-right, top-to-bottom) 
is at no advantage over any other.

The catch is that the numerical workspace ordering determines the 
correspondence between a horizontally-oriented pager and a 
vertically-oriented pager.  We'd like (I think) the spatial 
relationships between workspaces to be preserved as much as possible 
between the two pager orientations.  This is important if a user uses a 
vertical and horizontal simultaneously, or alternates between the two.

According to this reasoning, there are precisely two criteria/objectives 
in designing a workspace numbering scheme: A) that the numbers increase 
from 1 to N on 1D virtual desktops (easy); and B) that the horizontal 
and vertical orientations are spatially close to each other.

Based on these, I think proposal C (from my previous post) is the best 
of the lot:

Horizontal
1 2 3 4
5 6 7 8

Vertical
5 1
6 2
7 3
8 4

With this design, the vertical pager layout is a 90-degree clockwise 
rotation of the horizontal pager layout.  As far as I can figure, there 
is no orientation that better preserves the spatial relationship between 
a horizontal and vertical pager layout.


*Counter Arguments:

1) It might be true that some users would like to navigate a 2-d desktop 
via absolute numbers in addition to / instead of a directional 
navigation system.  So the criteria for choosing an ordering pattern 
given above might have been too strict.  Even so, I think that making 
vertical and horizontal pagers correspond spatially is probably a more 
important goal.

2) Maybe it would be better not to rotate the workspace at all?  For 
instance, I bet that a lot of users will be using a 2x2 (square) or 3x3 
(also square) desktop.  Rotating these layouts doesn't make any sense, 
and is detrimental from any perspective (numerical addressing OR spatial 
correspondence).  Let's look at Seth's recent poll to see how many users 
have 4 or 9 workspaces:

# workspaces |  % respondents
------------------------------
          1  | 14
          2  |  4
          3  | 24
          4  | 38
          5  |  6
          6  |  5
          7  |  1
          8  |  4
          9  |  1
         10  |  1
         11  |  1

More people have 4 than any other number.  But we don't know if these 
workspaces are arranged in a square or a sequence...

While it wouldn't (most likely) be worth the programming effort to 
special-case square layouts to keep them from being rotated on vertical 
panels, I can see how one might make the argument that layouts should 
*never* be rotated, period.  But the consensus seems to be that rotation 
is a good thing overall, and I tend to agree.

Overall, I'm happy with layout C.  What do you guys think?

-Michael