Re: [ns] strange "invalid command" error (bug?) when simulating

[Haobo Yu <haoboy@isi.edu>]

Are you recycling/reusing those tcp agents? I've seen a couple of these
problems before, and all of them were because I deleted those objects in
some place but used them in somewhere else. I tracked it down by adding
"puts" in the otcl files where it deletes those objects (if you know they
are only deleted in otcl) and the object name matches what it complained
about. Then you'll have an idea where/why it disappeared before you used
it again. Just my 2 cents.

- Haobo

On Mon, 17 Apr 2000, Richard Mortier wrote:

> Hi,
> 
> I'm trying to rerun a set of simulations that I previously had
> working, as far as I'm aware using the same versions of NS &
> associated libraries, compilers, CPUs, etc, etc.  However, the
> simulation frequently produces the following error (after
> approx. 150s simulated time):
> 
> """
> 
> ns: _o6912 produce 17.0: invalid command name "_o6912"
>     while executing
> "_o6912 produce 17.0"
> 
> """
> 
> ...with the _oxxxx value being different for different values of
> various simulation parameters and seeds.  If I try and print out
> "[ info class ]" on that object, it appears to be an "Agent/TCP"
> in the same way that the previously created ~6000 (in this case)
> were.  I'm not running out of memory, or other resources as far
> as I can see.
> 
> I'm using NS 2.1b4 (I'm trying to avoid porting some code to
> 2.1b6 at the moment), Otcl 1.0a3, tclcl 1.0b8 and tcl/tk8.0p2
> which I believe are all interoperating versions.  gcc/g++ 2.8.1
> is being used to build on a RedHat 5.2/Linux 2.2.2 machine.
> 
> So -- anyone any ideas as to what might be causing this?  It's
> probably something obvious that I'm missing, but I'm really
> stumped at the moment (and given that it takes about 10mins. to
> reproduce it each time, debugging it is going to be slow :-( ).
> The error is being printed as a result of calling
> init.tcl:unknown, if that's any help.
> 
> Cheers,
> 
> --
> Richard Mortier                 --------------------
> -----------------        ------ Computer Laboratory,
> rmm1002@cam.ac.uk  ----- University of Cambridge, UK
> mort@ieee.org ---- http://www.cl.cam.ac.uk/~rmm1002/
> 
>