Lego RCX-morph and SQLego stability

Edmund Ronald eronald at cmapx.polytechnique.fr
Thu Aug 15 10:30:37 UTC 2002


I am using a Mac OS 9.2 on a TI 400Mhz PB, with a Keyspan serial adapter,
a Lego serial tower (bought part) with a Lego Din to Serial cable, and an
RCX 2.0. Your basic stuff works ok - can download small programs and
execute commands like beep. However, there seems to be some forked reading
process which sits there and eventually times out ??? I am in an
electronics store in Tokyo now, will go home, and post the details if I
reproduce the bug. But basically, I would like to wrap all the code in
error-handlers so that I can run some robotics training routines that work
without my intervention, recovering from comms failure gracefully ...

Edmund
PS. The above does not relate to Ned's morph not loading ... However if
you have fixes for the comms routines, please tell me!



On Thu, 15 Aug 2002, Bergel Alexandre wrote:

> Hello Edmund,
> 
> Some pratical questions :
> Which OS do you use ?
> Which release of LegoMindstorms do you use ? Actually, there is *no* support for USB, so, only system below 2.0 could be used.
> 
> Alexandre
> 
> 
> On Wed, Aug 14, 2002 at 06:45:52AM +0200, Edmund Ronald wrote:
> > Hi Ned and others interested in Lego + Squeak -
> > 
> >  I cannot manage to load in Ned's  RCXMorph file, with or without
> > Alexandre's classes loaded into a 3.2 Squeak. I was interested in Ned's
> > code as Alexandre's stuff seems to work *sometimes* and often puts up an
> > error notifier.
> > 
> >  Error notifiers are OK for interactive work but a nuisance
> > if you are using a learning algorithm ...
> > 
> > Anyway, my questions are 
> > 1. Is there a *stable* version of the comms routines for the RCX ? How do
> > I debug the error notifiers I get intermittently ?
> > 2. How do I figure out the loading problems for RCXMorph ? I am a newbie
> > albeit a middle-aged one :)
> > 3. What is the state of play on Lego and Squeak ? Shall we form a SIG ? :)
> > 4. How does one work around error notifiers at run-time - ie catch the
> > notifier with a different process and fail gracefully - without having to
> > rewrite complex existing code like the RCX  comms stuff ?
> > 
> > 
> > Edmund
> > 
> > 
> > 
> 
> 




More information about the Squeak-dev mailing list