Re: Tracefile starts at random row

[Lee Breslau <breslau@research.att.com>]

Martin,

By design, each source picks a random starting place in the trace
file.  This is so that sources are not synchronized.

If you want to explicitly make them synchronized, the change
is very simple.  The starting place is picked by the following
statement in traffictrace.cc in the method TraceFile::setup():

	/* pick a random starting place in the trace file */
	return (int(Random::uniform((double)nrec_)+.5));

You could change that to:

	return (0); 

to accomplish what you want.

Lee

Date: Mon, 14 Feb 2000 12:24:16 +0100 (03:24 PST)
From: Czermin Martin <czermin@dune.LKN.E-Technik.TU-Muenchen.DE>
To: "ns-users@mash.CS.Berkeley.EDU" <ns-users@mash.CS.Berkeley.EDU>
Subject: Tracefile starts at random row

Hi all,

I try to run a simulation with real tracefiles  from an video
conference.
The starting time for each trace is set at 0.0000,

$ns at 0.0000  "$source($i) start"

but ns only starts the first trace at that time with the first row. The
other traces start, as far as I see, at 0.0000 but not on the first row.
They start with a different row for each trace.
But to get worst case conditions I must start them at the same time and
same row to get an estimation of the possible delay.
The thing confusing me is that ns start the different traces always at
the the same position (row) of the traces, which means for example
always in the 150th row of the tracefile.

If anybody know how I can force ns to start all the traces with the
first row , please be so
kind and send me a hint or a explanation of this behaviour.

Thanks in advance

Martin