On May 3, 2007, at 9:13 AM, Christian Theune wrote:


Am Donnerstag, den 03.05.2007, 08:59 -0400 schrieb Fred Drake:
On 5/3/07, Christian Theune <[EMAIL PROTECTED]> wrote:
In a small side note: I still don't know how we recommend using the eggs in a way that says "I want the egg-version of Zope 3.4.0" and the users gets those eggs (and only those) that are synchronised satellites and
have "3.4.0" as their version number.

I don't think we've defined what "the egg-version of Zope 3.4.0" means
(nor do I think we need to as part of a classic zpkg-based Zope

The 'and have "3.4.0" as their version number' I consider a non-goal.

Maybe a meta-egg that would introduce the dependency of "zope.xx==3.4.0"
as their dependency would allow that to happen because if this gets
involved in a working set then the synchronized satellites would be
restricted to the exact version.

Are you suggesting the satellites would have this dependency?  If so,
-1.  If not, and this is just used as a way of collecting the eggs, I
think this isn't valuable, and hence is a decoy.

I might have stated my goals the wrong way. I find it valuable to be
able to predict which exact versions of things "get pulled in" from a

The current way that dependencies are declared is that when I run
buildout in a year I'll get the zope.app.publication-3.5dev-r122 egg
on a stable application. I don't want that. I want to again and again
and again get the 3.4 egg of zope.app.publication because those eggs
where tested together.

OTOH. I'm just remembering that Jim talked about some `freeze` feature
for buildout ... is that what I seem to want to tackle this issue? ;)


http://www.python.org/pypi/zc.buildout/1.0.0b23#repeatable-buildouts- controlling-eggs-used


Jim Fulton                      mailto:[EMAIL PROTECTED]                Python 
CTO                             (540) 361-1714                  
Zope Corporation        http://www.zope.com             http://www.zope.org

Zope3-dev mailing list
Unsub: http://mail.zope.org/mailman/options/zope3-dev/archive%40mail-archive.com

Reply via email to