agronholm / anyio / 24424972435

93%

push

github

fix: prevent TaskGroup re-entry with clear RuntimeError (#1110)

Entering a TaskGroup context multiple times caused an AttributeError because __aexit__ deletes self._exceptions in its finally block. On the second __aenter__, the attribute is never recreated.

Fix: check if the TaskGroup is already active in __aenter__ and raise RuntimeError instead of allowing re-entry, which would produce confusing errors. Also reinitialize _exceptions for safety.

Fixes #1109.

Signed-off-by: bahtya <bahtyar153@qq.com>

* fix: add tests and changelog for TaskGroup re-entry guard

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix: add re-entry guard to trio TaskGroup backend too

Signed-off-by: bahtya <bahtyar153@qq.com>

* Apply suggestion from @agronholm

* Apply suggestion from @agronholm

* Apply review feedback: add blank line after raise in asyncio backend

* style: add blank line between changelog entries per review

Signed-off-by: bahtya <bahtyar153@qq.com>

* fix: correct re-entry test to reuse same TaskGroup

The test was creating a new TaskGroup instead of reusing the one
that was already entered, so it wasn't actually testing re-entry.

* Update src/anyio/_backends/_asyncio.py

* fix: remove duplicate raise, fix changelog author casing

* address review: remove dead code and fix test

Signed-off-by: bahtya <bahtyar153@qq.com>

* Remove TaskGroup._active flag, use _entered for lifecycle tracking

The _active flag was redundant with _entered. Since _entered now prevents
re-entry, _active is no longer needed. The spawn check in asyncio backend
uses cancel_scope._active instead.

As requested by maintainer in review comments.

Signed-off-by: bahtya <bahtyar153@qq.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix: restore _active flag for trio TaskGroup spawn check

The _entered flag alone is insufficient for the trio backend's spawn
... (continued)

9 of 9 new or added lines in 2 files covered. (100.0%)

6018 of 6488 relevant lines covered (92.76%)

10.1 hits per line