[Zope] Report on Solaris Zope performance available

[Oliver Erlewein]

Hi Tony

I saw the report yesterday and read it immediately. It is an EXCELLENT
summary of whats been discussed the last year or so! The explanations are VERY
understanable and allow for easy catching up on the subject. I'm very glad that
the report has been written.

At this point (although it's a little late - diddn't know that you were
writing up a document) I'd like to offer my time as review partner if there is
something more coming on threading. 

There are some points in the document that I don't quite agree to. I'll
adress these as soon as I have time for it (I'll have to do that on my private
time).

This morning I reconfigured my testcomputer with the infos from the report
and I still have the same problem. Could you add the exact compile options for
Python and the setup information for Zope?

I tried this morning to compile with SUN Forte (cc). Solaris 2.8-sparc,
Python 2.1.2:

export LD_LIBRARY_PATH=/usr/lib/lwp
export CC='cc -Xc -mt -lpthread -R /usr/lib/lwp'
./configure --with-threads --without-gcc --with-cxx=CC
make
make install

There were sone errors that came up on the *Splitter* files. There you just
have to delete some unimportant comments (or ^M's) and then all goes well. 

"Recompiled" Zope (2.4.4b1) with python wo_pcgi.py

I still have the "single process" thread. I'll have another go with gcc
today or tomorrow. I'll post the results. I'll compile with:

export LD_LIBRARY_PATH=/usr/lib/lwp
./configure --with-threads
make
make install

****
What the decision for or against Zope is concerned I don't quite understand
the problem. The question here is Solaris against Linux, FreeBSD,...
The Zope yes/no question is something completely different. The chain of
argunents used for the decision makers must be clear on that! These TWO subjects
only mingle when the company/varsity is fixed on Solaris as only OS.
****

Regards Oliver Erlewein

-- 
GMX - Die Kommunikationsplattform im Internet.
http://www.gmx.net