It causes occasional deadlocks, resulting in false negatives in Travis.
Disable the test for now. Works around #6540.
It causes occasional deadlocks, resulting in false negatives in Travis.
Disable the test for now.
Works around #6540.
ACK. We don't use multiple thread scheduling anyway, I think?
ACK-- we use a single scheduler thread.
Then why not remove/rewrite the code instead of putting a comment around it?
ACK
Then why not remove/rewrite the code instead of putting a comment around it?
This is a temporary measure (that's why it says "for now" everywhere). Obviously actually fixing the code would be preferable, but in all the time #6540 was open this has not been done - so seemingly non-trivial? Right now in the bustle before the 0.12 branch we don't want to have Travis false negatives interfering.
(edit: I tried to cut this Gordian knot of threads in #6171, but apparently there is still another problem than the one solved there)