Re: not reading variables

[Paul Johnson]

Stacy Oerder wrote:
> Hi
>
> I'm experiencing a problem reading in global variables(integers) into my
> swarm from my .scm file.
I'm afraid some of the expert C programmers will see you say "global 
variables" and go berserk because global variables are bad.  But I don't 
think you mean global in the C sense, you mean parameters in your grid.

> Once I've compiled my program and run it, it is
> apparent that my variables set in the .scm file are read correctly into
> my ModelSwarm file but they do not read in correctly in my bug and
> foodspace files. 
That seems odd to me, because when I have a mistake in the scm file, it 
is generally causing a segmentation fault.

I suspect the problem might be this. If you are working from a standard 
application like heatbugs, then that code has the command to read in the 
modelswarm parameters from the file. But when you create the foodspace, 
you need to read that in as well.  But I bet you don't have a line which 
instructs that object to be loaded from a file. I mean something such as:

 id archiver = [LispArchiver create: [self getZone] setPath: 
"my-lisp-file"];
 foodspace = [archiver getObject: "foodspace"];
 [archiver drop];

(fill in my-lisp-file with the name of your file).
Notice I'm not relying on any global objects to take care of my 
archiving because I don't know what all might be available.  It is easy 
enough to create an archiver, have it load your foodspace, and kill it off.

Every object you want loaded has to be explicitly loaded in this way. 
Your Observer loads the modelswarm, the model probably has to load the 
others.

Now, if you run your model and put in commands to save the objects, such 
as, you will see what you need.  This one creates the file name on the 
fly, but you can just put in some string and cut that. (It also talks to 
my parameters object, but you don't need that).  If you put something 
like this in your model swarm and then cause it to run at some point, it 
will create the *scm file and let  you see what is in there.

This one here does a shallow save on the model (shallow means save only 
ints, doubles, and other variables that are actually inside the model) 
but it does a deep save on my agent list (citizenList). Deep save will 
try to save objects within objects within objects. It can get really 
complicated, but it does work as long as all objects in side there have 
lispOutDeep: methods that make it work.

- lispArchive: (char *)inputName
{
  char dataArchiveName[100];
  if (!inputName)
    snprintf(dataArchiveName,100,"%s%d-%s.scm","run",[parameters 
getRunArg],"blah");
  else
    snprintf(dataArchiveName,100,"%s%d-%s.scm","run",[parameters 
getRunArg],inputName);
  id dataArchiver = [LispArchiver create: [self getZone] setPath: 
dataArchiveName];

  [dataArchiver putShallow: "model" object: self];
  [dataArchiver putDeep: "citizenList" object: citizenList];

  [dataArchiver putShallow: "parameters" object: parameters];

  [dataArchiver sync];
  [dataArchiver drop];

  return self;
}

I'm only bringin this up because saving a foodspace will generate a 
pretty massive file.  It will save not only the original value of the 
food you were talking about last week, it will also try to save the 
entire lattice of current food values.  I doubt your current *.scm file 
will have all that in there already. So I suggest having your program 
write an scm file so you can see what is supposed to be in there.



> Oddly enough the one variable does read into the
> foodspace, namely the amount of original food, but the world size does
> not read in correctly. And again at my bug file the world size reads in
> correctly but the value for the original food does not get read in. I
> constantly get the value 0 for all other variables, that is the food to
> be added as well as the world size for the two different files. Does
> anyone have an idea why this seems to be happening? I am a newbie with
> SWARM and am still coming to terms with the code.
>
> Thanks in advance.
> Stacy Oerder
>
>                   ==================================
>    Swarm-Support is for discussion of the technical details of the day
>    to day usage of Swarm.  For list administration needs (esp.
>    [un]subscribing), please send a message to <address@hidden>
>    with "help" in the body of the message.



--
Paul E. Johnson                       email: address@hidden
Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66045                FAX: (785) 864-5700


                 ==================================
  Swarm-Support is for discussion of the technical details of the day
  to day usage of Swarm.  For list administration needs (esp.
  [un]subscribing), please send a message to <address@hidden>
  with "help" in the body of the message.