[ns] About processing track files

["Nikolaos Katsarakis Austauschstud. (M. Schweigel)" <katsarak@ifn.et.tu-dresden.de>]

Dear ns-users,

I am doing some simulations and processing afterwards the track files,
e.g. to find packet delays.
(Using slightly modified version of L. Wood's script in
http://www.ee.surrey.ac.uk/Personal/L.Wood/)

For a quite small track file ~12 MB, ns-2 runs for about a minute and
the script for about 2 minutes.
As the next simulations are going to create much bigger files
(~50-100MB), I would like to ask:

1) Is it faster to record the track file to hard disk and then
post-process it?

or

2) Is it faster to pipe the track file directly to the post-proccessor?

Disk space is not a problem. According to the time() command, they are
equally fast. Shouldn't the first one be faster?

If somebody has tested it before, please drop me a line.

Thank you very much

Nick Katsarakis

PS: I also tried using perl (converting the awk script with a2p), but it
was actually slower.