[sugar] Sugar on the DCON chip

[Federico Mena Quintero]

On Mon, 2006-07-10 at 22:36 +0200, Marco Pesenti Gritti wrote:

> I think Diana design is excellent (also everyone likes it!), so I think 
> it should be our target. Though remember that we are still iterating 
> over sugar design. Diana is working on mockups for the "Home" page 
> introduced by design review 3... We have a meeting specifically targeted 
> at visual design at the end of July... Things could change heavily.

End of July is pretty late for the Summer of Code ;)

Attached is a tiny program to simulate the "swizzling" that is done in
the laptop's display.  It doesn't support the kind of interpolation that
the DCON chip can do, but it will give you a good idea of how things
will look.  Run it like "swizzle in.png out.png".

The results, when viewed at 1:1 on a normal display, will look terrible:
dark, murky, and fuzzy in a weird way.  But they will look different on
the DCON display, as it has a higher resolution than your screen and it
is way brighter.  Still, with a bit of imagination you can begin to see
what the problems are.

I think Diana's design is gorgeous, and it just needs some fattening of
lines and elements here and there.  We'll also have to see what font
looks good on the display; probably one that is big enough and has fat,
non-curvy strokes.

> We need to deal with technical limitations... Do you or Manu have a good 
> idea of what are going to be the problems? Or is there any documentation 
> about it? Certainly being able to emulate the display in xephyr would be 
> a big help... 

Yes, definitely.  Manu, as soon as you have the theme engine more or
less working in the torturer, please work on making Xephyr use the
swizzling code.  The stuff you need to modify is around
xserver/hw/kdrive/ephyr/hostx.c:hostx_paint_rect().  That's the function
that blasts the RGB buffer to the X server.  Mail me if you need more
details on how to do it, or how the swizzling works in general.

Attached is Jim Gettys's mail with a detailed description of how the
DCON chip works.

> > - We need write access to the GIT repository, so that Manu can put his
> > patches in a branch (or whatever the GIT terminology is for that).
> > Could someone please help us with this?
> >   
> I just sent mail to Ivan about this, he is usually very responsive.

Fantastic, thanks.

Manu, learn how to make branches in GIT or whatever they call them
there :)

> - Cleanup current code. It was put together quickly for the demo and 
> it's certainly messy, incomplete, buggy in several places.

OK.  Manu tells me that it fails the very first test in his torturer, so
that's a good place to start :)

> - There are lots of details in Diana mockups (shadows for examples) that 
> are just not implemented in current code or that are wrong. It should be 
> easy to figure out those by just comparing the mockups with current Sugar.

Shadows may not look very good on the DCON display.  It has high
resolution, but low acutance.  So, you can't display very fine details
or very subtle gradations with good visual results.  Anyway, run the
swizzle program and see how it affects things.

Thanks for the quick response!

  Federico
-------------- next part --------------
A non-text attachment was scrubbed...
Name: swizzle.c
Type: text/x-csrc
Size: 2434 bytes
Desc: not available
Url : http://mailman.laptop.org/pipermail/sugar/attachments/20060710/5d0290f3/swizzle-0001.bin
-------------- next part --------------
An embedded message was scrubbed...
From: Jim Gettys <jg at laptop.org>
Subject: Re: simulation of dcon chip
Date: Mon, 12 Jun 2006 15:13:19 -0400
Size: 11153
Url: http://mailman.laptop.org/pipermail/sugar/attachments/20060710/5d0290f3/attachment-0001.mht