It's time to start thinking about how best to transition in support for python 3 in gentoolkit. Pkgcore currently supports py 2.4~3.1 with a single code base by dynamically detecting the current python version and running 2to3 over the codebase if necessary. Portage also needs only a single code base, but relies on backported features, so can only support >=py2.6. We currently support >=2.5 to <3. If we're interested in keeping 2.5 support for some time, ferringb's 2to3 method will be necessary. If no one is interested in maintaining 2.5 support, then using the backported features will be the easier solution.
I see no reason for requiring 2.5 support so lets support >=python-2.6
Is anyone already working on this? If not it sounds like something fun that fits the "do something more productive in the evenings then reading half of the cheezburger network" bill ;) More seriously though, I did some preliminary experiments and it seems that doing a 2to3 conversion and then fixing up the result to work flawlessly with 2.6 and 3.x seems like a quite workable way to move forward with this (I guess most of the work will be in thoroughly testing everything)
Andy, No one is working on the conversion yet. Since gentoolkit svn access is restricted to devs, most of the "community" work on gentoolkit right now is happening at genscripts.googlecode.com. FuzzyRay, gentoolkit's maintainer, tracks genscripts trunk and pulls it into gentoolkit svn every few weeks. Send an email to me and tell me which email addy you prefer for google services and I'll give you commit and branching perms. We'll cook up a py3k branch.
*** Bug 307691 has been marked as a duplicate of this bug. ***
For those of you who are interested, gentoolkit-9999 contains support for python3. As soon as my workload slows down at my real life job, it will also be included in gentoolkit-0.3.0_rc10.
This bug has been fixed in app-portage/gentoolkit-0.3.0_rc10, thanks for reporting this issue.
