[Csci551-talk] TTL in project B

[John Heidemann]

On Sun, 20 Apr 2003 22:19:40 PDT, siddharth vora wrote: 
>
> Hi,
>
> My question is, If every node maintain the number of packets sent by a particular SENDER we can detect the loop.
>
>
> But, in this project we are not going to transfer from Shared tree to Source specific tree so, I think we have to implement TTL to prevent the loop.
>
>Please correct me if I am wrong !!

This has only shared trees and doesn't require nodes to keep track of
packets set.

So I would encourage folks to implement TTLs as the specification says.

   -John Heidemann

>
>Siddharth.
>2632 Ellendle St.,
>Appt. # 117
>LA, CA 90007
>Phone : 323 737 7522
>
>----- Original Message -----
>From: John Heidemann <johnh@ISI.EDU>
>Date: Sunday, April 20, 2003 9:49 pm
>Subject: Re: [Csci551-talk] TTL in project B
>
>> On Sun, 20 Apr 2003 11:01:11 PDT, jigesh vora wrote: 
>> >Hi,
>> >  since we are using only muticast trees for communication there will not 
>> be any loops. So can someone tell me the use of TTL in the messages?
>> >
>> >Thanx
>> >Jigesh
>> 
>> You're correct that correct code will not have any loops.
>> 
>> All students in the class who can guarantee that all their code will
>> work completely correctly the first time are not required to implement
>> TTLs.
>> 
>> For the rest of the class, you might find that TTLs prevent
>> catestrophic meltdowns if you accidently get a loop while you're
>> testing your code.  (Think about it: what happens if you accidently
>> get a loop where the packets get sent to multiple places... every time
>> they loop back.)
>> 
>>   -John Heidemann
>>