[Zope-Coders] *tap *tap* is this thing on?

[Guido van Rossum]

> On Thu, 2003-04-03 at 02:11, Richard Jones wrote:
> > On Thursday, April 3, 2003, at 02:47 PM, Richard Jones wrote:
> > > I've got a couple of minor changes to make in the installation 
> > > (runzeo.py isn't installed and the bin script command-line help output 
> > > isn't useful), and wanted to make sure I'm not stepping on any toes.
> > 
> > OK, having now looked into the ZEO situation in 2.7 more closely, I've 
> > found all sorts of places that it's broken. Mostly this has to do with 
> > the installation method. I've checked in fixed mkzeoinstance and 
> > mkzeoinst.py to get most of the way there.

[ChrisM]
> Thanks!
> 
> > I've hit a nasty blocker - the zdrun daemon attempts to execute 
> > runzeo.py, but without setting the sys.path in some manner that allows 
> > runzeo.py to find ZConfig. That's a bugger to track down 'cos the error 
> > is eaten up during the brief time "off the air" between zdrun 
> > fork/exec'ing and runzeo setting up new logging. Gah.
> 
> One strategy would be to sed out a small shell script like the "zopectl"
> shell script (which is written to an instance_home's "bin" during
> install") which itself sets the PYTHONPATH and calls zdctl.
> 
> > Is this on anyone's plate?
> 
> Not that I know of.  But guido may have an opinion about this...

In my opinion, this is not on my plate.  mkzeoinst.py etc. were
written for use in the new ZEO, which assumes that ZEO and ZODB have
been installed in site-packages (or PYTHONPATH set to point to where
they are).  I understand that Zope has different requirements; as long
as the changes for Zope don't break the stand-alone behavior, I don't
care what is done.

--Guido van Rossum (home page: http://www.python.org/~guido/)