roland.illig%gmx.de@localhost writes: > Am 25.11.2015 14:39 schrieb Greg Troxel <gdt%ir.bbn.com@localhost>: >> >> This definitely should have been discussed... > > I'm already sorry for that. I discussed it in private with Thomas, > although probably not in the depth necessary for such a change. > > Sure I can import the old pkglint, or I could revert the whole > thing. Or maybe improve the bootstrapping situation by importing gccgo, > if that would help. I think importing the old version is adequate. I can certainly see the point that the new implementation (especially having tests) has advantages. Turning go in pkgsrc into a portable language would be great, and not just for this, but also syncthing. It's clear that people are writing more and more things in go, and my only problem with it is that it's currently more like ghc than python, in that ghc is a hassle or impossible in many situations, and python builds on any system with C and POSIX, more or less. As long as the go version doesn't trivially build everywhere, I think we will have to set our expectations that commits be pkglint-clean to be only with the perl version. But certainly people can run the new one and fix things that our docs say are wrong.
Attachment:
signature.asc
Description: PGP signature