This: - Fixes the picked-version bug - Suppresses spurious (and not spurious) legacy version warnings. (When that setuptools/packaging problem is fixed. I'll release 2.3.2 that undoes this.) If we're still in significant pain tomorrow, then I'll release a 2.4 that requires setuptools <8. This isn't a long-term, because old setuptools won't handle new-style (better) versions. Jim -- Jim Fulton http://www.linkedin.com/in/jimfulton
Jim Fulton schreef op 16-12-14 14:13:
This:
- Fixes the picked-version bug
- Suppresses spurious (and not spurious) legacy version warnings.
(When that setuptools/packaging problem is fixed. I'll release 2.3.2 that undoes this.)
If we're still in significant pain tomorrow, then I'll release a 2.4 that requires setuptools <8. This isn't a long-term, because old setuptools won't handle new-style (better) versions.
zc.buildout 2.3.1 seems good here, thanks. For the record, the spurious warnings while bootstrapping are still there, but zc.buildout cannot fix it there. Donald is working on that in setuptools/packaging. During the buildout run the warnings are not there. There is one thing that surprises me, but that may be nothing new. If you pin older versions of zc.buildout or setuptools and have newest=false in your buildout (or call 'bin/buildout -N') then these older versions are not used. Bootstrap gives you the most recent ones and the bin/buildout run will not upgrade/downgrade the versions. The '_maybe_upgrade' call in buildout returns immediately. Sample buildout config: [buildout] parts = test # uncomment this as you see fit: newest = false [test] recipe = zc.recipe.egg eggs = zest.releaser [versions] setuptools = 7.0 zc.buildout = 2.2.5 zest.releaser = 3.53.2 Anyway, newest=true is the default in zc.buildout. -- Maurits van Rees: http://maurits.vanrees.org/ Zest Software: http://zestsoftware.nl
participants (2)
-
Jim Fulton
-
Maurits van Rees