* Jan-Wijbrand Kolman <janwijbrand@gmail.com> [2010-10-06 13:39]:
On 10/6/10 12:49 PM, Wolfgang Schnerring wrote:
* Jan-Jaap Driessen<jdriessen@thehealthagency.com> [2010-10-05 18:09]:
Version 0.12.2 of z3c.recipe.compattest is not compatible with recent versions of it's dependencies zc.buildout (v1.5.1), zc.recipe.egg (v1.3.2) and z3c.recipe.scripts. I fixed this in revision 117253. Is it OK with you to drop compatibility with zc.buildout versions< 1.5 in the next release of z3c.recipe.compattest?
I'm all for moving forward, but I don't really understand why the 1.5-update should cause such compatibility breakage.
I think in this particular case z3c.recipe.compattest relied on zc.buildout internals that now have have changed. So, no, I do not think it really is an API change and as such it is not intentional breakage nor something that __should__ be fixed in buildout.
Ideally, z3c.recipe.compattest would be fixed in such a way that is does make use of official buildout APIs.
Thanks for the explanation! I agree with you, it would make a lot of sense to use official APIs and not work around them. Don't know how hard that's going to be in this case, though. ;)
Still there's the case that quite some recipes need to be updated to zc.buildout 1.5.x if they want to use its new features ("system python support").
Ah. I didn't remember that compattest indeed does generate scripts (I thought it was delegating everything, but that's not correct of course), and I didn't understand from Jan-Jaaps email that system-python-support was a goal here, too, I only got the "some compatibility broke" part and was confused about that.
BTW, I discussed updating the z3c.recipe.i18n for similar reasons - to support newer buildout features - in a separate thread. There people seemed to be +1 on releasing a recipe that declares its minimal buildout version requirement to be >= 1.5.1. Would you say that that situation is similar to this?
Yes, I agree, that makes sense, so +1 from me here. Wolfgang