[Zope-Perl] Re: Standalone ZODB for Perl outside of Zope

Michael Fair michael@daclubhouse.net
Mon, 28 Apr 2003 12:15:04 -0700


> There's still time to switch!  You can start working on your Python Zen
> and actually use ZODB <0.2 wink>.

I actually tried to get the whole project to move to Zope,
but the "complete rewrite" idea got shot down pretty quickly.

It's a Perl/CGI application and looks like it's going to stay
that way for a while longer.


> > There's a toolkit
> > called Spread that is designed to make these kinds of
> > group communications as efficient as possible while still
> > making gaurantees about message delivery that I would
> > probably use.
>
> I'm well aware of Spread, c.f. http://www.python.org/other/spread/.
> We use it for our Zope Replicated Storage product.
>
> Note that Spread would be useful for only a small piece of the ZODB
> problem -- a database involving multiple networked clients.  The hard
> problems I mentioned above still need to get solved in the single
> process model.

Ok - I haven't done it yet, so I'll trust you. :)
I assumed that the Spread Toolkit was really just going
to handle getting messages pssed around - nothing more -
nothing less.  I'd also need to read up on their "virtual
synchrony" techniques to ensure that a process can determine
which messages have been received and which haven't.

One thing I hadn't done, but probably could is use the
WebDAV mount client and just have Perl use the standard
filesystem...  I'd of course need to do more research...
But at that point I'd probably just be using some
distributed file system like CODA, NFSv4, CIFS, or Intermezzo.

As you can tell, I've got a lot of learning to do.

-- Michael --