Maxima servermode: stability of commands....

You might have to experiment with different lisps.
Experiments with Allegro Common Lisp 7.0 and a
"maxima" dll  on windows XP, on a 2.5GHz Pentium 4
 suggest that a second, third,
etc system loads up in a fraction of a second. Maybe 100-200 ms?
I don't know if GCL shares as much; CLISP might also be faster
than GCL on this.

You could try visiting the web site   telnet://maxima.franz.com
(see http://www.franz.com/success/demos/maxima/ )
and time it yourself.  or in a shell,  type  telnet  maxima.franz.com

RJF


Gerd Kortemeyer wrote:

> Hi All,
>
> On May 2, 2006, at 12:01 PM, Richard Fateman wrote:
>
>>
>> I think that starting up a second maxima, given that most of
>> the pages are shared, can be done very fast, if your lisp
>> supports this approach.
>
>
> Actually, it does take a long time. I did some experimenting for LON- 
> CAPA, also for assessment purposes, and given the workloads we  
> regularly deal with, the one-process-per-request method is prohibitive.
>
> I wrote some code that starts a configurable number of Maxima  
> processes and keeps them alive. You can connect through a local port.  
> You can find the code at
>
> http://www.lon-capa.org/cas.html
>
> Under the hood, this basically does screen harvesting. Every time a  
> new client connects, it does a "kill(all);reset();" It has time-out  
> code, whereupon it kills the child process and starts a new one, and  
> maintains a blacklist of commands.
>
> Unfortunately, just like Michele reports, this stopped working with  
> the most recent update of Maxima I installed. It's Finals Week here  
> at MSU, so I did not have the time yet to find out why ... and now  
> back to writing exams :-(
>
> - Gerd.
>