upstream header fixes

[thehesiod]

fix to #1300

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.84%. Comparing base (6437c89) to head (3f15b21).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1301      +/-   ##
==========================================
- Coverage   89.84%   89.84%   -0.01%     
==========================================
  Files          67       67              
  Lines        6087     6086       -1     
==========================================
- Hits         5469     5468       -1     
  Misses        618      618              

Flag	Coverage Δ
unittests	89.84% <100.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[thehesiod]

btw ran the local only tests as well and they passed

[thehesiod]

hmm, perhaps we should try publishing a dev version of this for people to test with?

[jakob-keller]

hmm, perhaps we should try publishing a dev version of this for people to test with?

Sounds reasonable, but I have no experience with it. Would you expect a significant amount of people to notice and participate?

[jakob-keller]

As expected, the arm64 job on Python 3.8 is still flaky, but since we marked it as experimental, that does not block the PR. You could merge it now, if you like.

[thehesiod]

hmm, perhaps we should try publishing a dev version of this for people to test with?

Sounds reasonable, but I have no experience with it. Would you expect a significant amount of people to notice and participate?

I'm thinking someone in that discussion group linked, had a post recently: #1300. Let me change this to a dev version instead then

[jakob-keller]

I'm thinking someone in that discussion group linked, had a post recently: #1300. Let me change this to a dev version instead then

They could get the wheel from https://github.com/aio-libs/aiobotocore/actions/runs/13508980065 without us having to publish to PyPI.

[jakob-keller]

Let me change this to a dev version instead then

Thinking about it, I don't think our CI will work for a dev release. For starters, the CI trigger would not even fire:

aiobotocore/.github/workflows/ci-cd.yml

Line 11 in 7cb07f6

[0-9]+.[0-9]+.[0-9]+

[thehesiod]

Let me change this to a dev version instead then

Thinking about it, I don't think our CI will work for a dev release. For starters, the CI trigger would not even fire:

aiobotocore/.github/workflows/ci-cd.yml

Line 11 in 7cb07f6

[0-9]+.[0-9]+.[0-9]+

fixed (theoretically)

[maresb]

I'm seeing a regression related to this change on some old code:

TypeError: object bytes can't be used in 'await' expression

triggered by

aiobotocore/aiobotocore/endpoint.py

Lines 54 to 55 in 3f15b21

	else:
	response_dict['body'] = await http_response.content

I'm having some trouble with my debugger at the moment, so I'll try to provide more info once I get the chance to look around.

[thehesiod]

@maresb that line did not change in this PR, did you mean to address a different PR?

[maresb]

That particular line has not changed for several years, but this PR modifies the code path immediately before it.

It seems that else is being used as an implicit fallback, and it's clearly not meant to handle bytes, but something leading up to that recently changed and is now allowing bytes in my particular case.

[thehesiod]

ok if you could provide a testcase that would be greatly appreciated, thanks!

[kjagiello]

I was able to reproduce it, but only when running aiobotocore under moto.

repro.py

import asyncio
from aiobotocore.session import get_session
from moto import mock_aws

session = get_session()


async def main():
    with mock_aws():
        async with session.create_client("ec2", region_name="eu-west-1") as client:
            await client.describe_instances()


asyncio.run(main())

Run with uv:

uvx --with aiobotocore==2.21.1 --with moto==5.1.1 -- python repro.py

[maresb]

Thanks @kjagiello that's a great MWE! The code I have also uses moto.