Report Generator

From: Alex Rousskov (rousskov@ircache.net)
Date: Mon Dec 27 1999 - 15:33:54 MST


Greetings,

        I am aware of these three ways Polygraph users analyse their
benchmarking results:

   1. check whether the experiment finished
   2. check whether the experiment finish with reasonable
      "executive summary" stats produced by lx
   3. study "executive summary" stats, plot various
      distributions and traces to understand the results

I was always terrified to find out that somebody is using the first
approach (and many users do!). Needless to say that this approach may
lead to complete misinterpretation of the results. Unfortunately, even
the second approach is not good enough. In our Lab we try to use the
third strategy, but even we are being bitten by lack of time and
resources and miss important details.

Clearly, Polygraph is at least partially responsible for users not being
able to analyse the results in a proper way. It is difficult for a
non-programmer to use lx output and automate the analysis.

Report Generator released with Poly 2.2.8 today is the first step to
address the problem. The tool still requires a lot of work, but it is
already capable of generating very useful reports, close to the ones we
are using internally in our Lab.

We encourage you to go through the pain of installing and studying a new
piece of software. Fairly complete documentation and sample reports are
available. As usual, please report bugs to polygraph-bugs@ircache.net.

Docs:
        http://polygraph.ircache.net/UserManual/reportgen.html
        http://polygraph.ircache.net/doc/reportgen.html
Samples:
        http://polygraph.ircache.net/UserManual/polyrep/

Ready to be flooded with more 'on vacation' responses,

Alex.



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:10 MDT