Hi, as promised I'm currently trying to write down the instructions for running ZTK tests. I need some feedback on our process. I figured there are various scenarios for which we have tools: 1. Ensure that a branch of the ZTK stays in working order For that the ZTK has test runners created by z3c.recipe.compattest which ensures that the individual tests of the ZTK packages with their respective dependencies using the specific ZTK versions are ok. This helps when evolving the versions in the ZTK. Those should be run nightly as is done by some of the buildbots. 2. Ensure that a change of a package doesn't break the set 2a: ensuring compatibility with a specific branch of the set The developer working on package X can use a compat test runner and a specific ZTK buildout configuration to run the other packages' tests combined with the development version of package X. This can/should be done by individual developers for the (newest/olders?) ZTK version they target their change for. If we can encode the "target ZTK" version for a specific package in that packages' configuration then we can also have nightly builds that perform this verification automatically. I'm not sure how to encode that, though, because the relation between the package and the ZTK is usually defined the other way around. 2b: ensuring ongoing compatibility of the trunks In addition I wonder how the various trunk versions of ZTK packages should relate to each other. Should they always build? Should we accept breakage of all trunks with each other or not? Also, there is not version of the ZTK (I think) that specifies 'use the trunk of everything', right? Christian -- Christian Theune · ct@gocept.com gocept gmbh & co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development