Re: bug in tcp.cc?

[Yuri Pryadkin <yuri@ISI.EDU>]

It shouldn't fail if you explicitly set it to 0, but it could fail if
you set it to an expression that evaluates to 0 (e.g. 1/3*3-1).  But
it doesn't do much harm, does it?  If overhead_ is a very small
positive number, the delsnd_timer_ will be rescheduled by an even
smaller value (half of it, on average).  Besides, how are you
suggesting to fix it?  If we use something like (overhead < EPSILON),
what EPSILON should we pick?

  Yuri.

Jitendra Padhye <jitu@cs.umass.edu> writes:

> Hi,
> 
> In tcp.cc: (ns version 2.1b2)
> 
> =====
>     if (overhead_ == 0 || force) {
>       output(t_seqno_++, reason);
>       npackets++;
> =====
> 
> However, "overhead_" is of type "double", so the equality test
> may fail even if overhead is set to 0 ...
> 
> - Jitu