Re: [Gzz] 9th (asko)

[Benja Fallenstein]

Asko Soukka wrote:

> So, before hurrying with the next deadline I
> would like to get to know with current client (changes from 0.6.X, where
> are all keybindings, views, dimlists...). 

There are three important views: row, column, and vanishing. And then, 
there's the xubuoy view from the demo which is currently less generally 
useful.

The key bindings are in gzz.view.DefaultBinder at this time. I'm 
thinking about re-writing that part to modularize.

The dimlists are kept in Fallback, but are not very useful :(, because 
there is no way to create new dimensions.

Changes from 0.6: A terrible lot ;-)

> Guess, there's still good stuff
> to convert from 0.6.X (eg. views)...

Your mind* views would be cool (and also, in related news, the 
ball-and-stick cell view). If you do re-implement them, please consider 
following Ted's rule that interpretation of a corner list should stop 
when another "left" connection is encountered:

  A - B
  |   |
  |   C
  |   |
  D - E
      |
      F

This structure should be interpreted as two corner lists: One with A as 
the header and B and C as the subordinate elements, and one with D as 
the header and E and F as the subordinate elements. The connection C-E 
is ignored (for the grouping of cells). So B/C/D/E would not be a circle 
in the mind* views: B/C would be one and E/F another.

The purpose is for example chronological stuff: A could be "January" and 
D "February," and B/C could be "January 1st" and "January 7th," and E/F 
"February 5th" and "February 12th." So you have a long chronological 
rank of days, and for nice grouping, a shorter chronological rank of 
months. When showing this in mind*, each month should become its own 
"wheel."

> Benja: Would it be too bad to discriminate JNI tests with different
> filename extension? As I recall, currently the extension is given already
> in the makefile.

Hm, it would be possible, but wouldn't it be better to discriminate them 
based on the directory, as I proposed earlier? I mean, we want the 
platform-specific code to be in special directories anyway for clarity...

- Benja