ns2.1b5 and CMU mobility code

["Samir R. Das" <samir@cs.utsa.edu>]

Hi all:

I continue to have problems understanding the how the new
CMU mobility code fits with rest of ns. Here are my problems:

1. Even with the new phy.cc (1.5) dsr fails at certain assertions.
   This is with the given test suite but with longer simulation time. 
   As I mentioned here before, short simulation time may not
   generate any traffic for this test.

   dsdv core dumps with some other traffic/mobility model imported
   from CMU site. 

   But my bigger problem is next. 

2. I played with the CMU code in the past (obtained from their web site)
   before this official release. They separate out the IP addr and port
   address in two different fields, while in ns they are clubbed
   together into one 32 bit number (nsaddr_t).

   Now in the new release, it seems to me that things got mixed up. For
   example, arp.cc assumes that the src/dst field in the ip header is pure IP
   address. Same for dsdv.cc. I am not entirely sure about dsr.cc (this
   was a little hard to follow). But the rest of the ns, it is the old
   ns style.

   Isn't it a problem? Well, I am facing some problems already in arp
   for the routing code I am writing. This is what I understand as the
   source of the problem. Can somebody clarify? I can point out 
   the actual locations of ns release, which I think may be problematic.  

Thanks,

Samir