[omniORB] Measuring omniorb's footprint. New footprint benchmark.

[Dan Kegel]

Craig Rodrigues wrote:
> > [http://www.kegel.com/dkcorbabench]
> I don't know if you are aware of this, but there is
> a CORBA Benchmarking web page:
> 
> http://nenya.ms.mff.cuni.cz/~bench/

Yes, I've read it.  (I link to it from 
http://www.kegel.com/corba/ )  They have some nice results.
 
> Are you collaborating with the person who put together
> that web site?

No.  dkcorbabench is something small I am doing for my own needs.
I need to find an orb for a limited-memory system, and to
help pick from the available alternatives, I wrote a very
simple benchmark.  The orb with the highest score on this
benchmark is probably the right orb for my project.

As a bonus, I'm writing it using autoconf, so a single
Makefile and a single set of sources will work for all orbs.
(Eventually.  Omniorb3 works now; I'll try Tao next.)
I expect the autoconf macros I'm using will be useful for
real projects.
 
> The OMG has a web site called Corba Open Source Testing,
> http://cost.omg.org.  However, I don't think that
> effort has really gone anywhere.

Worse, that site is offline.  According to archive.org,
their last update was in October, 2001.
Their download page said:
"  To view using regular web interface:
   http://cost.omg.org/cgi-bin/cvsweb.cgi/

   To access the CVS repository as an anonymous guest:
   export CVSROOT=:pserver:anonymous@cost.omg.org:/pub/cvsroot
   cvs login
   password is "guest"

   From now on you can execute all read-only commands, e.g.
   cvs checkout cost
   to checkout all COST sources."
Does anyone have a copy of those tests?

> Your numbers regarding footprint will be of special
> interest to people looking at embedded CORBA applications.

Thanks.  I'm hoping they'll be relevant.

- Dan