On Sat, Jul 14, 2001 at 04:51:11PM +0300, danielv at netvision.net.il wrote:
> Pretty cool.
> There seems to always be an arbitrary (surprising) elements to the way
> the constraints are resolved when I change things - is the solver trying
> to change as few variable as possible? 

Yes, this is the nature of the simplex algorithm that Cassowary is based
on.  

> This might make a Morphic/Cassowary interface somewhats strange to work
> in..

Possibly.  I imagine that one would get used to it.

> 
> Or maybe there's a way to specify what should change?

Well, you can tell some values to change in preference to others (eg:
change a morph's position before you change its extent).  However, if
the objective function can be optimized just as well by changing one 
variable a lot, Cassowary will never distribute the change over multiple
variables.

Thanks for your interest,
Joshua


> 
> "Joshua 'Schwa' Gargus" <schwa at cc.gatech.edu> wrote:
> > Hello,
> > 
> > I've finally solved a small but significant bug that was causing
> > several of the Cassowary demos to work improperly (it was my fault, 
> > not that of the fine folks at Washington).  Anyone who has played
> > with the  demos and thought "what the heck is that supposed to be
> > doing?" should try again; several of them have fundamentally
> > different behaviors.
> > 
> > Both the HTMR-Cassowary project on BSS and the semi-permanent
> > download location (http://penguin.cc.gt.atl.ga.us:8080/schwa/63)
> > have been updated.
> > 
> > Also in the works is a first attempt at integration into Morphic,
> > which should provide a more compelling demonstration than the 
> > small examples already provided.
> > 
> > Enjoy,
> > Schwa
>