What should stable NixOS prioritize?

Personally, I think there’s 2 avenues we could go:

Optimize for people’s time:

  • Move the release to be YY.05 and YY.11
  • Provides ample time for the latest gnome and plasma DM’s to be packaged. Also allows for the related ecosystems to also support those releases. Major distro’s which specialize in shipping a specific DM will likely have fixes for the issues we encounter
  • This essentially “aligns the release around unstable and external ecosystems”

Optimize for current YY.03 and YY.09 timeline:

  • any fundamental package which has many dimensions of compatibility (glibc, gcc, llvm) should be merged in with ample time to address regressions in unstable (e.g. a month). Otherwise they should be put off until after a branch off. (We saw 3 staging next PRs which added 500 staging commits a few days before the target 20.09 branch-off date; which did hurt the release schedule). Discussion around staging/staging-next convetions should be done in Fixing the staging/staging-next workflow
  • systemd will have to align with the version of gnome and plasma we will ship.
  • However, this will cause more divergence between the two branches (if the critical packages get updated after branchoff, then the two branches could exhibit different regressions), and ZHF may suffer.
  • This essentially “aligns unstable around the release”

Personally, I’m in favor of pushing our release back two months, so that the Desktop ecosystem is likely to be much more manageable. For production users, they will still have the 6 month cadence, past the initial push back.

12 Likes