Closed cstruct closed 3 years ago
Thanks again for the report @cstruct, I'm a bit horrified that the test suite didn't catch such an obvious mistake :sweat_smile:
I'll add a new test for that.
@cstruct I realized while working with the example that you provided that it actually shows not one but three different issues with aiostream, now fixed with PR #75.
The first one is related to a typo:
[working() in range(100)]
This evaluates to False
, which is not a valid source. This generated an exception after the failing()
sources had been scheduled. With #75, this kind of typo will generate an exception much sooner, when the stream is instantiated.
Thanks again!
Thank you for quick responses and fixes! With these fixes I can avoid a segfault in aioodbc
.
If there are still pending tasks in a
TaskGroup
when its__aexit__
is called theself._pending
set is modified during iteration.Causes: