swarm-support
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Swarm GUI


From: Theodore C. Belding
Subject: Re: Swarm GUI
Date: Fri, 9 Aug 1996 15:02:35 -0400

Hi everybody-
Regarding probes, there's an outstanding bug affecting HPs that prevents
probes from working at all.  (The Swarm team knows about this and is
working on it).  I'm obviously biased, since we're using HPs here at
Michigan, but I'd be in favor of this being fixed before any additional
bells and whistles are added. Unfortunately, it may be a problem with the
gcc Obj C runtime (as is the problem with PowerPC, POWER, and RS/6000
platforms -- I'm beginning to wonder whether the runtime isn't just a big
kludge). (A debugger that actually works with Obj C would also be very
useful. Of course, it's unclear how much of this the Swarm team should take
on itself. It's too bad that GNU Obj C isn't ready for prime time.)

As Sven pointed out earlier:
>An interesting question is, given limited programmer resources, should the
>Swarm
>team emphasize efficient batch operation or glitzy (well, not to slant
>this too
>much, let's say 'intuitively helpful') graphical representations? The
>answer is not
>necessarily obvious ...

My personal opinion is that the GUI is good enough for now.  For serious
science, batch runs are more important.  There are a few things that still
need to be implemented or fixed so that real batch runs are possible.  For
example, we need to be able to record the complete initial state of a run,
including all of the RNG seeds, so that the run can be duplicated later --
is this currently possible?  I've heard something about an internal RNG for
the scheduler or some such: is there a way to get the initial seed for that
RNG, if it exists?  I think Manor said there would be some improvements to
the batch mode in Swarm in the next release.

Regarding the limitation of only one object per cell in the 2-D spaces, the
fastest way to fix this would probably be for one of us to contribute a new
class.
-Ted

--
Ted Belding                        mailto:address@hidden
University of Michigan Program for the Study of Complex Systems
http://www-personal.engin.umich.edu/~streak/




reply via email to

[Prev in Thread] Current Thread [Next in Thread]