Jim Fulton wrote:
Adding a call to addsitedir won't accomplish your stated goal of dropping eggs into lib/python and having them get used. For addsitedir to have any impact, you have to also manage the .pth files.

I don't know what you mean here. I don't recall managing any .pth files... If I add addsitedir to the right places, the eggs will be picked up. I have a test egg for the Document Library here that pulls in pyoai, lxml (dependency of pyoai), hurry.query, zc.catalog (dependency of hurry.query), hurry.workflow, and hurry.file.

If I include the right ZCML in package-includes, Document Library works.

I hope to start using zc.table and its dependency (zc.resourcelibrary) eventually as well. It's nice I can do this with just a tweak to the Document Library's setup.py and a new ZCML include.

The fact that you "requently get lost between sys.path, sys.prefix, .pth files and the site.py module" is not at all surprising. There is a lot of magic there. We certainly don't want to toss any of this in at the last hour (I wish) of a release.

I'm just distributing a script that tosses this into any given Zope 3 instance now, so people can at least use this stuff in Zope 3.3. And yes, there's too much magic going on, and I hope zc.buildout will make life easier and less magic on the longer term.


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

Reply via email to