Re: static vars

[donalson]

Sorry,

   My last post wasn't clear, it was just a warning to Cygnus users.
The background is this:

I realized that I had a Swarm "movie player" if I just played my data
files back through my simulation (with some slight mods to my sim.)  In
fact it needed only one object, plus model and observer.  It also helped
solve a second problem.  I have a HUGH amount of data per sim run, one
file for each of 440 cells and each cell saves 61 variables per time step
for 3000 timesteps.  I use the data player to also make stripped down data
sets of particular variables I am investigating.

Agents can enter or leave cells to 20 nearest neighbors and I keep track
of where they come from and where they go.  Unfortunatly, rather stupidly,
I keep a running sum instead of just the number per time step.  The
obvious solution to this is to read in the array (size 20 ints) save it,
read in the same array for the next time step, subtract them, and then
save the new array as the old, and repeat.

I started by making "oldData[20]" static, but I discovered that the data
in it was getting changed from one iteration to the next.  I could verify
that it was correct just prior to exiting the method, but at next access,
it had changed.  It is not improbable that this is caued by mishandling of
static vars when there are multiple insances of the same object.  This is
just a guess but the incorrect data was all within the expected range of
the variables, just the incorrect for that instance.  I tried it with a
single variable and found the same problem.  I should note that these vars
are not accessed outside the object, and the variable names are used
nowhere else in the simulation.

Since I needed a persistent variable, my next thought was to move the
static variables to the object definition (removing the "static" part of
the definition).  The make consisted only of the data object (I had
changed nothing else) and the result was that the schedule did not run!
It went through initialization and put up the displayes and then just sat
there.

I finally got it to run correctly by doing a "make clean" prior to
recompile.  I think the moral of the story may be Cygnus has some trouble
with static variables.

Cheers,

   D3

Marcus Daniels wrote:

> >>>>> "D3" == donalson  <address@hidden> writes:
>
> D3> It is not happening.
>
> What is not happening?  You need to explain what you want and expect
> to happen, identify some common reference frame, and show us the
> suspect mechanics before we can possibly help.
>
> I guess you want to do some processing based on some vector state
> change, apparently chunking from some input file, but it is hard to
> tell with the information you've provided.
>
> D3>   Since you and I agree on that, I have to believe that there is
> D3> something weird in Cygnus.
>
> You mean contrasted against some other environment?
>
>                   ==================================
>    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.
>                   ==================================

--
*********************************************************************
* Doug Donalson                 Office: (805) 893-2962
* Ecology, Evolution,           Home:   (805) 961-4447
* and Marine Biology            email address@hidden
* UC Santa Barbara
* Santa Barbara Ca. 93106
*********************************************************************
*
*   The most exciting phrase to hear in science, the one that
*   heralds new discoveries, is not "EUREKA" (I have found it) but
*   "That's funny ...?"
*
*       Isaac Asimov
*
*********************************************************************



                  ==================================
   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.
                  ==================================