[meta-sig] persistence SIG

[Nicolas Chauvat]

> I'd really like participants not to focus too much into bringing one
> particular implementation into the standard lib.  It should be
> something like a common subset of required functionalities.

I think Jim Fulton did a good job at defining what were the components of
a persistence framework and how beneficial it would be to have these
components standardized separately.

<may-be-skipped>
My personal opinion is that a "persistence framework" is not the kind of
thing that you can sell to a client and what you can not sell to a client
is to be considered "raw material" or "commodities" and will benefit
everyone if it is stable and very low cost. Having it part of the standard
library or part of an "official persistence extension module" (e.g.
NumericPython or PIL) doesn't make much of a difference.

Same goes for OS, web server, etc. When your clients want a website, they
do not care whether it runs on Apache, Zope or something else. They want
the web site to fit their needs, look pretty and be fast and secure. If
providers can rely on Apache or Zope for security and speed, they'll
compete on prettiness and fitting clients' needs using skills and in-house
developed software: things that money can buy.
</may-be-skipped>

-- 
Nicolas Chauvat

http://www.logilab.com - "Mais oł est donc Ornicar ?" - LOGILAB, Paris (France)