https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=270358

--- Comment #223 from Ulas SAYGIN <[email protected]> ---
FWIW, I independently tested updating devel/py-setuptools to 83.0.0 (the
current
upstream release as of this writing) on a clean FreeBSD 15.1-RELEASE-p1 amd64
poudriere jail, before finding this existing PR.

Changes I made mirror what's already been discussed here: PORTVERSION bump,
removed the now-defunct pkgutil.ImpImporter patch (pkg_resources has been fully
removed from setuptools as of this version, so that patch target no longer
exists), and removed the BUILD_DEPENDS pin on py-wheel044<0.46.0 (83.0.0
vendors
its own wheel 0.46.3 under setuptools/_vendor/).

Testing performed:
- Baseline `poudriere testport` on unmodified 63.1.0: PASS
- `poudriere testport -p main devel/py-setuptools@py312` on updated 83.0.0:
PASS
  (build/stage/QA/install/deinstall all clean)
- portlint -A / -AC: 0 fatal errors (2 pre-existing false-positive warnings,
  unrelated to this change)
- As a real-world non-Python consumer check: built dns/powerdns end-to-end with
  the updated setuptools installed in the jail, including its full dependency
  chain from scratch (boost-libs, curl, mysql84-client, postgresql18-client,
  sqlite3, lmdb, lua54, libsodium). All backend modules (bind, pipe, mysql,
  pgsql, sqlite3, lmdb) built and staged cleanly. Confirmed via INDEX-15 that
  powerdns, nginx, dovecot, and postfix don't depend on py-setuptools directly,
  so this is more of a sanity check than expected risk area.

This is a small, single-machine data point compared to the exp-runs already
posted here. Diff attached for reference (attachment ATTACHMENT_ID_BURAYA).
Happy to help re-run tests against whatever the current candidate patch is.
Thanks to everyone who's been pushing this forward — following comment
#218/#221, sounds like things are close.


Attachment #273243

-- 
You are receiving this mail because:
You are on the CC list for the bug.

Reply via email to