Skip to content

tests: Fix races in the read-growing-file test

First: the buffer is pushed on the encoding filesink only after the checkpoints, leading to wrong ordering in expectations.

The second one is about EOS being posted on the bus before the stop action is executed, leading to the action never being executed. That action was plain useless as the default behavior of validate scenario is to internally create a "stop" action on EOS (unless disabled by user).

Should fix !937 (comment 723604) which I could reproduce after a few hundreds iteration here, with that patch I have been running 5k iterations without any problem.

Edited by Thibault Saunier

Merge request reports