John Marino <netbsd%marino.st@localhost> writes: > On 10/4/2012 18:59, Greg Troxel wrote: > I am glad you like the suggestion. I think we still don't see eye to > eye on this "migration" issue though. If somebody is using python2.6 > for their own purposes (meaning non-pkgsrc programs) then he's lost > nothing. You're clearly talking about a system with packages built > with python 2.6 and the system owner switching to a new pkgsrc branch > to update/add to his packages. He wouldn't have the option if we > stopped building python 2.6 packages (unless he set > PYTHON_VERSION_ACCEPTED I guess). I believe that risk comes with the > territory of mixing branches. I view pkgsrc users as being sourced based, so building with pkg_rr on a new branch is my normal, rather than using binary packages. I agree that branch mixing is a bad idea. I'm talking about updating to the new branch without having to 26/27 migrate. > However, this is a good compromise: > - python2.6 exists > - python2.6 packages are not built by default in bulk > - The possibility exists to built packages with python 2.6 from > source, but the packages are not guaranteed to build or function > correctly. > > Then the guy isn't "forced" to migrate, but it would be in his best > interest to do so. Sure, that's fine. I agree that people sort of should, but also think that often there are many other considerations.
Attachment:
pgp0VML60P0O7.pgp
Description: PGP signature