python312Packages.nose: fix build #325968
Conversation
| +# Adapted from CPython 3.11. | ||
| +# Licensed under the PSLv2. | ||
| +# https://github.com/python/cpython/blob/3.11/LICENSE |
There was a problem hiding this comment.
This is just a nit because the practice is really common, but:
2. Subject to the terms and conditions of this License Agreement, PSF hereby
grants Licensee a nonexclusive, royalty-free, world-wide license to reproduce,
analyze, test, perform and/or display publicly, prepare derivative works,
distribute, and otherwise use Python alone or in any derivative version,
provided, however, that PSF's License Agreement and PSF's notice of copyright,
i.e., "Copyright (c) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023 Python Software Foundation;
All Rights Reserved" are retained in Python alone or in any derivative version
prepared by Licensee.
3. In the event Licensee prepares a derivative work that is based on
or incorporates Python or any part thereof, and wants to make
the derivative work available to others as provided herein, then
Licensee hereby agrees to include in any such work a brief summary of
the changes made to Python.
Would you be willing to make the patch add the full upstream LICENSE file as LICENSE.cpython file, and adapt the comment similarly to the following?
# Adapted from the CPython 3.11 imp.py code.
# Copyright (c) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023 Python Software Foundation; All Rights Reserved
# Originally licensed under the PSLv2 (see LICENSE.cpython) and incorporated under the LGPL 2.1 (see lgpl.txt).
There was a problem hiding this comment.
Hopefully all is set on the pynose side from mdmintz/pynose#33 (comment). Let me know if anything else!
| # 2to3 is removed from Python 3.13, so always use Python 3.12 2to3 for now. | ||
| preBuild = lib.optionalString isPy3k '' | ||
| ${python312.pythonOnBuildForHost}/bin/2to3 -wn nose functional_tests unit_tests |
There was a problem hiding this comment.
I guess when 3.12 leaves Nixpkgs we’ll have to do one of:
- package
2to3independently of the Python interpreter; - run it once over the
nosecode and just vendor the results; or - remove
nose?
There was a problem hiding this comment.
Yeah, that's roughly my thoughts. My order of preferences for future actions: (I assume most people are in a similar frame of mind, but just to jot it down in detail.)
-
Most preferred would be getting rid of nose dependencies upstream. It's unmaintained and most users don't use very much of its functionality, so a migration should be easy. For some packages, I think this will just be a matter of getting the version in Nixpkgs updated. For some simpler stuff,
nose2pytestmay be enough to generate a patch that we can use + submit upstream. -
Failing that, if Lack of attribution and licence information for code derived from CPython mdmintz/pynose#33 gets resolved then maybe we can consider the pynose licensing issue resolved and resume usage of pynose as a drop-in replacement, pending other's approval. I'm sure there will be some hesitance here though and I'm not going to argue if the thought is we'd rather play it safe.
-
Failing that, we can continue to try to patch Nose for Python 3.14+. In an earlier version of this PR I did indeed include a 2to3 patch file, but it's rather large, so I think putting it in Nixpkgs preemptively is undesirable.
-
And failing that, my thought is we can remove
nose, disable tests on anything that continues to usenose, and eventually remove packages that depend onnosewhich are unmaintained.
If you think it is a good idea, I can try to add more context to that comment for the future. Specifically, we could outline the options for handling 2to3.
Regarding packaging 2to3 independently: maybe even better would be packaging the fissix tool by amyreese, as it essentially seems to be an improved 2to3. (Coincidentally, it is also apparently used by the nose2pytest tool.)
There was a problem hiding this comment.
Ideally by 2028 we've fixed the ecosystem enough, that nose is not relevant anymore. I wouldn't worry too much about this today.
There was a problem hiding this comment.
I didn’t realize that we kept Python releases for that many years. Agreed that it’s not worth thinking about in that case.
|
Things have gotten a bit scattered here.
|
|
Due to the rebuild count this should target the staging branch. |
|
I am also running a I feel more confident in this approach than #325935. The other patch seems incomplete too and I still feel more assured about the provenance of this one. |
|
Thanks, re-rebased. |
|
I forgot that things like Tensorflow would be in the build chain… It was at 78 failures after >1,400 successful builds and 239 builds left to finish, which seems good to me. I forgot to start the job in the builder in |
|
Since |
|
Thanks for all of the help everybody! I had no ideal what an ordeal I was in for when I set out to try to fix this. Fingers crossed, but hopefully we can close the book on this one for now. |
|
I am still facing issue on darwin-nix aarch64-darwin reference #326151 Dependence on … while evaluating derivation 'python3.12-powerline-2.8.3'
whose name attribute is located at /nix/store/77ijmdbv96rhqj2sr7aq0i5q8p16ii0d-source/pkgs/stdenv/generic/make-derivation.nix:334:7
… while evaluating attribute 'propagatedBuildInputs' of derivation 'python3.12-powerline-2.8.3'
at /nix/store/77ijmdbv96rhqj2sr7aq0i5q8p16ii0d-source/pkgs/stdenv/generic/make-derivation.nix:388:7:
387| depsHostHostPropagated = elemAt (elemAt propagatedDependencies 1) 0;
388| propagatedBuildInputs = elemAt (elemAt propagatedDependencies 1) 1;
| ^
389| depsTargetTargetPropagated = elemAt (elemAt propagatedDependencies 2) 0;
… while evaluating derivation 'python3.12-python-hglib-2.6.2'
whose name attribute is located at /nix/store/77ijmdbv96rhqj2sr7aq0i5q8p16ii0d-source/pkgs/stdenv/generic/make-derivation.nix:334:7
… while evaluating attribute 'nativeBuildInputs' of derivation 'python3.12-python-hglib-2.6.2'
at /nix/store/77ijmdbv96rhqj2sr7aq0i5q8p16ii0d-source/pkgs/stdenv/generic/make-derivation.nix:378:7:
377| depsBuildBuild = elemAt (elemAt dependencies 0) 0;
378| nativeBuildInputs = elemAt (elemAt dependencies 0) 1;
| ^
379| depsBuildTarget = elemAt (elemAt dependencies 0) 2;
error: nose-1.3.7 not supported for interpreter python3.12 |
|
This change has to pass through staging first. Check back in two weeks. https://github.com/NixOS/rfcs/blob/master/rfcs/0026-staging-workflow.md |
|
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/error-nose-1-3-7-not-supported-for-interpreter-python3-12/48703/40 |
Description of changes
This is an alternative to #325935 which uses a patch that I made independently without looking at the other patches. If we can't figure that out, then this is a backup plan.
I anticipate that my patch is incomplete, but it is complete enough to pass checks on all of the packages that I just re-enabled checks for.
This also fixes Python 3.13 support in Nose. Nose relies on 2to3, which is removed in Python 3.13, so we always use 3to2 from Python 3.12. This should buy us some more time to wait for upstreams to either get off of Nose or get removed for being unmaintained.
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.