Package Details: python-anthropic 1.12.1-1

Git Clone URL: https://aur.archlinux.org/python-anthropic.git (read-only, click to copy)
Package Base: python-anthropic
Description: Python library that provides convenient access to the Anthropic REST API
Upstream URL: https://github.com/anthropics/anthropic-sdk-python
Licenses: MIT
Submitter: carsme
Maintainer: Auerhuhn
Last Packager: Auerhuhn
Votes: 5
Popularity: 1.13
First Submitted: 2024-03-23 18:05 (UTC)
Last Updated: 2026-10-08 08:33 (UTC)

Latest Comments

1 2 Next › Last »

Auerhuhn commented on 2026-10-03 21:17 (UTC)

@qubeck Thanks for the pointer. Going to retry the build.

qubeck commented on 2026-10-03 17:07 (UTC)

python-httpx2 is already 2.13.0-1. Could you update this now?

Auerhuhn commented on 2026-08-29 18:10 (UTC)

Updated to 0.125.0.

All 1.x releases are currently blocked by python-httpx2, which needs to be bumped to v2.9.0 or higher for python-anthropic v1.x to work.

Maxr commented on 2026-08-29 15:52 (UTC)

Please update or orphan, is outdated for weeks now.

bytemixer commented on 2026-07-30 08:34 (UTC) (edited on 2026-07-30 08:35 (UTC) by bytemixer)

Build fails on my machines, giving this error:

tests/lib/test_bedrock.py:254: in test_region_infer_from_profile
    assert client.aws_region == profiles[0]["region"]
E   AssertionError: assert 'us-east-1' == 'us-east-2'
E     
E     - us-east-2
E     ?         ^
E     + us-east-1
E     ?         ^
--------------------------------------------------------------------------- Captured log call ----------------------------------------------------------------------------
WARNING  anthropic.lib.bedrock._client:_client.py:87 No AWS region specified, defaulting to us-east-1
============= short test summary info ===============
FAILED tests/lib/test_bedrock.py::test_region_infer_from_profile - AssertionError: assert 'us-east-1' == 'us-east-2'
============= 1 failed, 1194 passed, 2 skipped, 1 xfailed in 12.62s ===============
==> ERROR: A failure occurred in check().
    Aborting...
 -> error making: python-anthropic-exit status 4

tijko commented on 2026-06-03 22:34 (UTC)

Getting failed to install exit status (8)

Auerhuhn commented on 2026-04-27 12:40 (UTC)

I have adopted the package, bumped the version and fixed the outstanding issues.

Auerhuhn commented on 2026-03-19 18:04 (UTC)

Hi @envolution,

thanks for maintaining this PKGBUILD.

Would you mind bumping it to the latest stable release?
Yours is twelve upstream releases behind, blocking other packages such as python-llm-anthropic.

Thanks!
Claudia

envolution commented on 2025-10-21 17:29 (UTC)

@sashank I've included your suggestion to remove the filterwarnings block, however the flags you suggested are unrecognised by pytest

ERROR: usage: pytest [options] [file_or_dir] [file_or_dir] [...]
pytest: error: unrecognized arguments: --benchmark-skip

I've installed pytest-benchmark and re-ran checks without issue

sashank commented on 2025-10-17 15:53 (UTC)

Hey @envolution,

So I think python-pytest-benchmark is required by CUDA python amongst other packages related to this whole AI stuff we're all working with, so its hard to uninstall it or python-pytest-xdist (which I tried for while, but this package forced me install it again).

So it seems like its an ongoing issue in regards to the conflict between benchmark and xdist: https://github.com/ionelmc/pytest-benchmark/issues/286

The above hints that we should look at the pytest config, which we'll find in pyproject.toml

What do you think of the idea of patching pyproject.toml, particularly this section?:

[tool.pytest.ini_options]
testpaths = ["tests"]
addopts = "--tb=short -n auto"
xfail_strict = true
asyncio_mode = "auto"
asyncio_default_fixture_loop_scope = "session"
filterwarnings = [
  "error"
]

So the first idea is to remove the lines filterwarnings = [ "error" ].

Another is that we could modify the addopts line to : `addopts = "--benchmark-disable --tb=short -n auto"'

I get that flag from the benchmark changelog: https://pytest-benchmark.readthedocs.io/en/latest/changelog.html#b2-2015-10-17

Another flag we could try is --benchmark-skip?

And then we can ask upstream anthropics folk to do the same?

What are your thoughts?

1 2 Next › Last »