[Sugar-devel] Screen Reposition?

Gary C Martin gary at garycmartin.com
Mon Aug 3 22:46:32 EDT 2009


Hi Art,

On 4 Aug 2009, at 02:42, Art Hunkins wrote:

> At:
> http://dev.sugarlabs.org/ticket/39
> I found the following observation:
>
> "I think that the right solution for such a tiny display is to  
> globally
> scale the display, using the bin/sugar to set SUGAR_SCALING to 72. The
> question is, how to detect display resolution from this shell  
> script? The
> only solution I can think of is an evil hack involving xrandr and  
> sed, and
> there has to be a better way."

Historically, all Activities were designed for the XO display at  
1200x900 with a scaling value of 100. Now that Sugar is running on a  
whole range of screens this has complicated layout somewhat. Setting  
scaling to 72 would be ideal for an 864x553, giving a widget layout  
that is equivalent to an XO display. The scaling setting as I  
understand it is something picked up from the GNOME world, 72 and 100  
are the only two supported values. Scaling does not affect font sizes  
(that's another whole hot potato).

> I'd like to experiment with setting SUGAR_SCALING, but cannot locate
> bin/sugar on SoaS. Can anyone give me the appropriate file/location?

It's an environment variable (open a terminal and type echo  
$SUGAR_SCALING to see what value it inherited). Where it's set depends  
somewhat on the distro, but I last saw it was in ~/.Xclients

> My idea is that if this variable works to "resize the screen", maybe  
> this
> value might be temporarily "reset" by a python script.

I'm not convinced that would be a wise thing to try ;-)

Sounds like the fun you've been having is in creating a flexible GTK +  
layout, or with some widget that wasn't designed to scale. Sorry I  
haven't been able to give direct feedback on your code posted so far  
but I'm rather backlogged at the moment (and am no great expert  
myself, I'd just have to go poke my way through the GTK+ API docs  
again).

Regards,
--Gary



More information about the Sugar-devel mailing list