Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

staging-next 2022-12-09 #205270

Merged
merged 264 commits into from
Dec 17, 2022
Merged

staging-next 2022-12-09 #205270

merged 264 commits into from
Dec 17, 2022

Conversation

raboof and others added 30 commits November 24, 2022 16:18
I don't think anyone sane is really using kernel < 3.10 nowadays.
(At least in a use case with glibc from nixpkgs.)

As another suspect with old kernels, Ubuntu seems to start at 3.13 now.
The .cpp file released with this package to PyPI was built using an
older version of Cython that is incompatible with Python 3.11. This
package is not actively maintained, so it's unclear when the fix for
this will be merged and released.

In the meantime, we can include a newer version of Cython locally,
which will be used instead to regenerate the .cpp file.
…jekyll-setup

jekyll-favicon: Add to full Jekyll setup
python310Packages.cffi: patch closures to work on M1 machines
python27: switch to ActiveState's fork for Python 2
…om intel-graphic-compiler version 1.0.12260.1, reduce platform to unix
Broken, unsupported, unmaintained. Constrained to EOL python versions.
Only required on python36, which was dropped a while ago.
Only required on python36, which was dropped a while ago.
Python 3.6 was dropped a while ago.
Python 3.6 was dropped a while ago.
Leaf package that doesn't build on supported python versions.
Backport of collections.abc additions from one EOL version to another.
@vcunat
Copy link
Member Author

vcunat commented Dec 16, 2022

I don't think many packages depend on these, but let me ping the maintainers anyway:

(uh, quite many people)

@SuperSandro2000
Copy link
Member

#206460

@vcunat vcunat merged commit 0ba3487 into master Dec 17, 2022
@vcunat
Copy link
Member Author

vcunat commented Dec 17, 2022

Also I let this regression pass to master:

@sternenseemann
Copy link
Member

clisp was fixed in #206745.

@dbaynard
Copy link
Contributor

I'd thought the httpie tests might be flaky, but I think the issue is with nixpkgs? That said, my local test outputs differ slightly from hydra's.

I can build the python3Packages.httpie from 5dc7114 but not from 04f574a — yet they use the same source. I have not checked which dependencies have changed.

The KeyError: 0 business is something I'd expect to see if there had been a change in the testing environment. Has there been some change with how nixpkgs python tests work?

The progress of failed tests (on darwin)
tests/test_auth.py ..........                                            [  0%]
tests/test_compress.py .......                                           [  1%]
tests/test_downloads.py ......                                           [  2%]
tests/test_errors.py ....                                                [  2%]
tests/test_httpie.py ...........s...............                         [  5%]
tests/test_json.py .                                                     [  5%]
tests/test_auth.py ..........                                            [  6%]
tests/test_compress.py .......                                           [  7%]
tests/test_downloads.py ......                                           [  7%]
tests/test_errors.py ....                                                [  8%]
tests/test_httpie.py ...........s...............                         [ 10%]
tests/test_json.py .                                                     [ 10%]
tests/test_auth.py .........                                             [ 11%]
tests/test_auth_plugins.py ....                                          [ 12%]
tests/test_binary.py ......                                              [ 12%]
tests/test_cli.py ......................................                 [ 16%]
tests/test_cli_ui.py ....                                                [ 16%]
tests/test_cli_utils.py ..                                               [ 17%]
tests/test_compress.py ..                                                [ 17%]
tests/test_config.py ........s                                           [ 18%]
tests/test_cookie.py .                                                   [ 18%]
tests/test_cookie_on_redirects.py .....................                  [ 20%]
tests/test_defaults.py .................                                 [ 22%]
tests/test_downloads.py ....................                             [ 24%]
tests/test_encoding.py .................FFFFFFFFFFFFFFFFFFFFFFFFFFFFFF.. [ 28%]
....                                                                     [ 29%]
tests/test_errors.py ....                                                [ 29%]
tests/test_exit_status.py .........                                      [ 30%]
tests/test_httpie.py ......................                              [ 32%]
tests/test_httpie_cli.py .....................................           [ 36%]
tests/test_json.py FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF [ 41%]
FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF [ 48%]
FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF [ 56%]
FFFFFFFFFFFFFFFFFFFFFFFFFFFF............................................ [ 63%]
.......................                                                  [ 65%]
tests/test_meta.py ......                                                [ 66%]
tests/test_offline.py .........                                          [ 67%]
tests/test_output.py ..........xXx...................................... [ 72%]
.......................................................................F [ 79%]
F                                                                        [ 79%]
tests/test_parser_schema.py .                                            [ 79%]
tests/test_redirects.py ...x......                                       [ 80%]
tests/test_regressions.py ...                                            [ 80%]
tests/test_sessions.py ................................................. [ 85%]
............                                                             [ 86%]
tests/test_ssl.py .ss.....pytest-httpbin server hit an exception serving request: TLS/SSL connection has been closed (EOF) (_ssl.c:997)
attempting to ignore so the rest of the tests can run
...............                                [ 89%]
tests/test_stream.py ..FFFFFFFF......                                    [ 90%]
tests/test_tokens.py ...................                                 [ 92%]
tests/test_transport_plugin.py .                                         [ 92%]
tests/test_update_warnings.py ...........                                [ 93%]
tests/test_uploads.py ........................                           [ 96%]
tests/test_windows.py s.                                                 [ 96%]
tests/test_xml.py FFFFFFFFFFFFFFFFF                                      [ 98%]
tests/utils/matching/test_matching.py ....................               [100%]

@vcunat
Copy link
Member Author

vcunat commented Dec 20, 2022

If you'll have no ideas, I could attempt bisection, at the cost of lots of CPU time.

@dbaynard
Copy link
Contributor

I have no ideas but I'm not familiar with the recent changes to the ecosystem. It's plausible another of the pythonPackages maintainers will know, right away — though none has responded here.

I don't mind doing a bisect, myself, as I'm doing some non-coding work, today.

@dbaynard
Copy link
Contributor

I see what you mean by lots of CPU time… the bisect points to a commit in this PR. Now I'm building everything, it seems (currently building llvm).

I'm ok with letting it run, but perhaps one of the python maintainers might chip in with a suggestion? I don't know who is most likely to know. I did skim the commits, but nothing obvious jumped out.

@vcunat
Copy link
Member Author

vcunat commented Dec 20, 2022

Yes, it's building from scratch, often a few times, with later bisect steps usually rebuilding much less.

@dbaynard
Copy link
Contributor

Just looking at the filres that have changes, the only thing that jumps out is removal of the (python) requests patch that changes certificate handling.

I've paused the bisect, for the moment, and instead reverted 8456141 to see whether It's the requests/certifi change that has caused the issue. There are other commits that refer to certifi; if this revert fixes the problem, then it's plausible the certificate handling in certify is incorrect.

I actually don't care about httpie that much, but I figured I'd try to be a good citizen 🤣 no good deed goes unpunished, etc. I'll ping when I have an update.

@mweinelt
Copy link
Member

c2068e0fff4d10a63ea46ee2db5917de8b9f7095 is the first bad commit
commit c2068e0fff4d10a63ea46ee2db5917de8b9f7095
Author: Fabian Affolter <[email protected]>
Date:   Sat Nov 12 19:06:39 2022 +0100

    python310Packages.responses: 0.21.0 -> 0.22.0

 pkgs/development/python-modules/responses/default.nix | 18 ++++++++++++++----
 1 file changed, 14 insertions(+), 4 deletions(-)

@dbaynard
Copy link
Contributor

dbaynard commented Dec 20, 2022

If I revert that, there is no error from httpie.

Test summary after reverting c2068e0
tests/test_auth.py ..........                                            [  0%]
tests/test_compress.py .......                                           [  1%]
tests/test_downloads.py ......                                           [  2%]
tests/test_errors.py ....                                                [  2%]
tests/test_httpie.py ...........s...............                         [  5%]
tests/test_json.py .                                                     [  5%]
tests/test_auth.py ..........                                            [  6%]
tests/test_compress.py .......                                           [  7%]
tests/test_downloads.py ......                                           [  7%]
tests/test_errors.py ....                                                [  8%]
tests/test_httpie.py ...........s...............                         [ 10%]
tests/test_json.py .                                                     [ 10%]
tests/test_auth.py .........                                             [ 11%]
tests/test_auth_plugins.py ....                                          [ 12%]
tests/test_binary.py ......                                              [ 12%]
tests/test_cli.py ......................................                 [ 16%]
tests/test_cli_ui.py ....                                                [ 16%]
tests/test_cli_utils.py ..                                               [ 17%]
tests/test_compress.py ..                                                [ 17%]
tests/test_config.py ........s                                           [ 18%]
tests/test_cookie.py .                                                   [ 18%]
tests/test_cookie_on_redirects.py .....................                  [ 20%]
tests/test_defaults.py .................                                 [ 22%]
tests/test_downloads.py ....................                             [ 24%]
tests/test_encoding.py ................................................. [ 28%]
....                                                                     [ 29%]
tests/test_errors.py ....                                                [ 29%]
tests/test_exit_status.py .........                                      [ 30%]
tests/test_httpie.py ......................                              [ 32%]
tests/test_httpie_cli.py .....................................           [ 36%]
tests/test_json.py ..................................................... [ 41%]
........................................................................ [ 48%]
........................................................................ [ 56%]
........................................................................ [ 63%]
.......................                                                  [ 65%]
tests/test_meta.py ......                                                [ 66%]
tests/test_offline.py .........                                          [ 67%]
tests/test_output.py ..........xXx...................................... [ 72%]
........................................................................ [ 79%]
.                                                                        [ 79%]
tests/test_parser_schema.py .                                            [ 79%]
tests/test_redirects.py ...x......                                       [ 80%]
tests/test_regressions.py ...                                            [ 80%]
tests/test_sessions.py ................................................. [ 85%]
............                                                             [ 86%]
tests/test_ssl.py .ss....................                                [ 89%]
tests/test_stream.py ................                                    [ 90%]
tests/test_tokens.py ...................                                 [ 92%]
tests/test_transport_plugin.py .                                         [ 92%]
tests/test_update_warnings.py ...........                                [ 93%]
tests/test_uploads.py ........................                           [ 96%]
tests/test_windows.py s.                                                 [ 96%]
tests/test_xml.py .................                                      [ 98%]
tests/utils/matching/test_matching.py ....................               [100%]

@alexshpilkin
Copy link
Member

alexshpilkin commented Dec 21, 2022

httpie uses requests, an HTTP library, and its tests rely on responses, which is a mock library for it. So it makes at least some sense that the test could have started failing when there were no changes to httpie itself, and I can hope that the failure is limited to the tests and the tool still works.

Just from looking at the logs, I suspect that getsentry/responses#585 (in 0.22.0).

Here’s one test failure from Hydra:

self = <httpie.models.HTTPResponse object at 0x7fffeebbe410>

    @property
    def headers(self):
        try:
            raw = self._orig.raw
            if getattr(raw, '_original_response', None):
                raw_version = raw._original_response.version
            else:
                raw_version = raw.version
        except AttributeError:
            # Assume HTTP/1.1
            raw_version = 11
>       version = {
            9: '0.9',
            10: '1.0',
            11: '1.1',
            20: '2.0',
        }[raw_version]
E       KeyError: 0

httpie/models.py:82: KeyError
----------------------------- Captured stderr call -----------------------------

http: error: KeyError: 0

There, self._orig is a requests.Response normally and a responses.Response in tests, self._orig.raw is a urllib3.HTTPResponse, and self._orig.raw._original_response is an http.client.Response normally, a responses.OriginalResponseShim in tests before that PR and a urllib3.HTTPResponse after it.

As best as I can tell, responses.OriginalResponseShim did not have a version attribute at all while urllib3.HTTPResponse defaults to (nonsensical) version=0.

I should probably go and check if that’s what actually happening, then report this upstream (to httpie or responses, I’m not sure). But if someone wants to pick that up I’m fine with it as well.

TL;DR: Probably an upstream bug / new version incompatibility limited to the tests.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
1.severity: security Issues which raise a security issue, or PRs that fix one 6.topic: golang 6.topic: kernel The Linux kernel 6.topic: nixos Issues or PRs affecting NixOS modules, or package usability issues specific to NixOS 6.topic: ocaml 6.topic: printing 6.topic: python 8.has: clean-up 8.has: module (update) This PR changes an existing module in `nixos/` 8.has: package (new) This PR adds a new package 10.rebuild-darwin: 501+ 10.rebuild-darwin: 5001+ 10.rebuild-darwin-stdenv This PR causes stdenv to rebuild 10.rebuild-linux: 501+ 10.rebuild-linux: 5001+ 10.rebuild-linux-stdenv This PR causes stdenv to rebuild
Projects
None yet
Development

Successfully merging this pull request may close these issues.