Eli Schwartz <eschwa...@gentoo.org> writes: > On 8/27/24 5:03 PM, Robin H. Johnson wrote: >> There wasn't an introduction message to this series, but I wanted to >> raise the discussion. >> >> We only JUST got rid of the last EAPI6 ebuilds in the main tree. >> >> There are overlays that still have EAPI6 ebuilds - and depend on these >> ebuilds. >> >> When is an expected time for all of those ebuilds to migrate, and how is >> that being communicated? > > > If we were removing an eclass that only supports EAPI 6 and is being > dropped because it's useless, we'd last rite it and give people 30 days > to move. > > But because the *file* isn't being removed, there is no rule how to do > it apparently?? :D The obvious answer here is to stick an ewarn in the > "if EAPI 6" branch at global scope. > > > (It's a bit messy when doing dependency calculation. This too is a > feature, if you think about it.)
Yes, it's something which has bothered me for a while. When we ratified GLEP 83 [0], I wanted to come back to it for handling EAPI support deprecation in "important" eclasses but I couldn't figure out a nice definition for that and got distracted. I actually *do* think we should do something here, but I will note pkgcheck will have been warning about use of DeprecatedEapi at least. [0] https://www.gentoo.org/glep/glep-0083.html
