Tres Seaver wrote:
But that code doesn't improve the non-purging mode. The changes Wichert
proposed make sense with or without the 'upgrade steps' feature.
If we had the upgrade machinery in place, we could scrap non-purging
mode altogether -- its purpose is to allow for "controlled" application
of changes to existing configuration without full replacement.
Wichert's rationale was specifically:
The particular reason I'm interested in this is that for Plone we are
playing with using GenricSetup profiles as part of the migration code.
This means we can write a lot less python code but just write changes
as profile snippets.
Well. If we don't use the non-purging mode we can't write changes as
profile snippets. Should upgrade steps always be implemented in pure
Python without using any XML files?
BTW: Are there any unit tests for the upgrade steps feature?
CMF 2.1 beta has some serious site manager issues. Please let's focus on
resolving these issues first.
I'm really just lobbying to have the GS work tested and merged. Fixing
ths LSM stuff is in the hands of a different set of folks, I think.
Rob, how did stuff go at Sorrento?
The CMF 2.1 branch is not as stable as it should be. Adding the 'upgrade
steps' feature might be low risk, but the other changes on the sprint
branch look more risky to me. I'm a bit afraid merging them will
destabilize the 2.1 branch further.
Zope-CMF maillist - Zope-CMF@lists.zope.org
See http://collector.zope.org/CMF for bug reports and feature requests