search

mudler / luet

:package: :whale: 0-dependency Container-based Package Manager using SAT solver and QLearning
https://luet.io
GNU General Public License v3.0
254 stars 24 forks source link

Unexpected atom gets into package #293

Closed msdobrescu closed 2 years ago

msdobrescu commented 2 years ago

Luet version:

In the log referred below.

CPU architecture, OS, and Version:

GH's.

Describe the bug

The configuration to be built here. The only dependency in the repo is here. In the build log, here dev-python/pybind11 atom is emerged. This is part of the dependency atom dev-python/pikepdf.

Why dev-python/pybind11 gets into the final package?

Generally, even though those deps of the direct deps of a package do not get there. Am I doing something wrong?

To Reproduce

Every time the package version is bumped and the package is rebuilt.

Expected behavior

Nothing related to dev-python/pybind11 should be included into the artifact/package.

Logs

Additional context

Other packages were built successfully like this. Is it an intended behavior change lately?

mudler commented 2 years ago

I can't access the build logs, but I suspect something must be going on between in the template so it ends up in the diff. Can you tell if downgrading luet gives a different result? I'm pretty sure nothing changed in this area and this is well covered by our test suite

msdobrescu commented 2 years ago

log.txt

mudler commented 2 years ago

I'm confused, from the logs it looks like dev-python/pybind11 gets emerged inside steps, so it would end up in the package indeed. If you want to strip it off, you can exclude path/files directly with excludes, or emerge -C

msdobrescu commented 2 years ago

Got it! Should not work like that! That package should be in pre_reqs actually! My bad! Not a bug. Thank you for this and sorry!