From 15000 kms away, both tests use a server running in a separate thread. Both servers certainly may write into the log file.
I don't know if such writes are atomics as far as python thread handling is concerned.
If they aren't, this sounds like yet another fallout from the thread/socket leaks.
A reliable way to reproduce this bug will be welcome but... if it's related to the leaks,
I won't hold my breath.
From 15000 kms away, both tests use a server running in a separate thread. Both servers certainly may write into the log file.
I don't know if such writes are atomics as far as python thread handling is concerned.
If they aren't, this sounds like yet another fallout from the thread/socket leaks.
A reliable way to reproduce this bug will be welcome but... if it's related to the leaks,
I won't hold my breath.