Re: node-field-size on >=ns 2.1b3, >128 nodes

[Padmaparna Haldar <haldar@ISI.EDU>]

On Fri, 24 Jul 1998, Lloyd Wood wrote:

> Finally compiled ns 2.1b3 under Solaris, and, as a check for what I
> report below, the current snapshot (23 July 1998).  Both exhibit the
> same behaviour. Running with tcl/tk 8.0p2, otcl 1.0a3, tclcl 1.0b6 -
> all mod cons.
> 
> Even though my scripts specify
> 
> Node expandaddr
> 
> I'm seeing:
> 
> Number of nodes exceeds node-field-size of 7 bits
>     while executing
> "error "Number of nodes exceeds node-field-size of $nodebits_ bits""
>     (procedure "_o3" line 4)
>     (Simulator check-node-num line 4)
>     invoked from within
> "$self check-node-num"
>     (procedure "_o3" line 8)
>     (Simulator node line 8)
>     invoked from within
> "$ns node"
>     (file "/user/ccsrnrpg1/eep1lw/ns/lloyd/torus6.tcl" line 210)
> 
> with scripts setting up large networks for both 2.1b3 and the current 
> snapshot. These scripts execute without complaint under ns 2.1b2.

"Node expandaddr" should be called before you start creating nodes.
could you check if that's the case and let me know if it still fails..

Node expandaddr has an equivalent command :
"$ns set-address-format expanded" that increases address size from 16 to
32 bits (basically the same thing that expandaddr does).

It's a part of ns's new address-format series API's.
Write-up for this is available in ~ns/tcl/lib/ns-address.tcl
But we still have Node expandaddr around for backward compatibility.



> 
> It looks like the new node-field-size code (which seems like a good
> idea) in ns-lib.tcl is somehow getting the wrong idea about how many
> nodes can be allocated. 
> 
> I'd expect a complaint about node-field-size of *8* bits rather than
> 7; 

1 bits is allocated for mcast (though this seems like a waste this is done
to take care of, again, backward-compatibility for the existing scripts),
so that leaves 7 bits for the nodes and 8 for ports, bu default.

--Padma.

>since I'm expanding the address space and using far less than 2**22
> nodes, I was rather surprised to see this.
> 
> Any other workarounds other than dropping back to 2.1b2?
> 
> thanks,
> 
> L.
> 
> <http://www.ee.surrey.ac.uk/Personal/L.Wood/>PGP<L.Wood@surrey.ac.uk>
> 
> 
> 
> 

------------------------------------------------
Be true to your work, your word, and your friend.
--Thoreau

Padmaparna Haldar
ISI/USC.
310.822.1511 #352