test(python-sdk): drop httpcore-era stream reader tests after the pyqwest migration

[mishushakov]

tests/test_file_stream_reader.py was written against httpcore and never migrated with the rest of the pyqwest stack — it builds bare httpx.Client() instances, so it still passes green while exercising a transport the SDK no longer ships. Both of its load-bearing premises are dead: _active_connections() read client._transport._pool.connections (an httpcore-only internal — PyqwestTransport has no _pool), and request.extensions["timeout"]["read"] is no longer a per-chunk idle bound, since the pyqwest adapter collapses read/write into one whole-operation deadline and exits the timeout scope before the body streams — verified directly on PyqwestTransport, where a 0.5s mid-body stall with read=0.05 raises no ReadTimeout. This deletes the five tests that asserted only httpcore behavior (both idle-timeout tests, the slow-consumer test, both abandoned-reader tests) plus the helper, and re-anchors the remaining eight on response.is_closed — FileStreamReader.close()'s actual contract, transport-agnostic and stronger than the pool check, since the context-manager tests now also assert the response stays open mid-stream. Also removes tests/bugs/, whose sole file was a permanently @pytest.mark.skip'd pyautogui repro against the desktop template.

The real streaming-idle coverage against actual pyqwest transports already lives in tests/test_volume_client.py; the SDK stopped sending per-request timeouts on streamed reads for this same reason in e2b/sandbox_sync/filesystem/filesystem.py:207-215.

Test-only, so no changeset — matching the repo convention for test(...) PRs.

Verification: tests/test_file_stream_reader.py 8 passed (was 13); full python-sdk unit suite 245 passed; ruff format, ruff check and ty check clean.

Closes SDK-324

🤖 Generated with Claude Code

[linear-code]

SDK-324

[changeset-bot]

⚠️ No Changeset found

Latest commit: b0a7bbb

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[cursor]

PR Summary

Low Risk
No production code changes; only test deletions and assertion refactors for the Python SDK.

Overview
Test-only cleanup after the pyqwest migration: stream-reader unit tests no longer depend on httpcore connection-pool internals or transport-level read timeouts that the shipped stack does not provide.

test_file_stream_reader.py drops five cases (sync/async idle timeout via request.extensions, slow-consumer parity, abandoned-reader-on-client-close) plus _active_connections and optional read-timeout wiring in _open_stream. The eight remaining tests assert response.is_closed after full consume, context exit, idempotent close, and async idle_timeout on AsyncFileStreamReader—matching the reader’s public lifecycle contract. Module docs and the chunked-server helper are updated accordingly (stall_before defaults to -1).

Removes tests/bugs/test_envelope_decode.py, a permanently skipped desktop/pyautogui envelope repro.

^{Reviewed by Cursor Bugbot for commit b0a7bbb. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Package Artifacts

Built from 08dc55b. Download artifacts from this workflow run.

JS SDK (e2b@2.38.4-sydney.0):

npm install ./e2b-2.38.4-sydney.0.tgz

CLI (@e2b/cli@2.16.2-sydney.0):

npm install ./e2b-cli-2.16.2-sydney.0.tgz

Python SDK (e2b==2.38.0+sydney):

pip install ./e2b-2.38.0+sydney-py3-none-any.whl

[claude]

LGTM — test-only cleanup, verified against the actual reader contract.

Extended reasoning...

Overview

Test-only PR touching two files under packages/python-sdk/tests/. It deletes 5 httpcore-era tests (and their _active_connections/request.extensions["timeout"] helper plumbing) from test_file_stream_reader.py that asserted behavior specific to the old httpcore transport, re-anchors the remaining 8 tests on response.is_closed, and removes a dead, permanently-skipped tests/bugs/test_envelope_decode.py file. No production code is touched.

Security risks

None. No production code paths, auth, crypto, or permissions are affected.

Level of scrutiny

Low. This is a test-suite cleanup with no runtime behavior change. I verified the core claims directly against the source rather than taking them at face value: FileStreamReader.close()/AsyncFileStreamReader.aclose() in e2b/sandbox/filesystem/filesystem.py call response.close()/response.aclose() respectively, so asserting response.is_closed is indeed the reader's real contract and a faithful (arguably stronger) replacement for the old pool-internals check. I also confirmed that real idle-timeout coverage against the actual pyqwest-backed transport already exists in tests/test_volume_client.py (e.g. test_sync_stream_stall_raises_read_timeout, test_async_explicit_stream_idle_timeout_aborts_stall), so the deleted tests aren't a net loss of coverage for that behavior.

Other factors

The removed tests/bugs/test_envelope_decode.py was unconditionally @pytest.mark.skip'd and exercised an unrelated desktop/pyautogui template, so its removal has no coverage impact. Per CLAUDE.md, test-only PRs don't require a changeset, which matches this PR. No unresolved reviewer comments in the timeline (only bot summaries), and the bug hunting system found no issues.