I think a race here explains the intermittent test failures that recently became more prominent.
I suspect it also fixes the memory leak observed by @dannyvanhoof in #49 . My theory is that if unsubscribe is called too soon, the subscription is still created after the component is unmounted and left dangling in memory. I have not done thorough testing of this though.
I think a race here explains the intermittent test failures that recently became more prominent.
I suspect it also fixes the memory leak observed by @dannyvanhoof in #49 . My theory is that if
unsubscribeis called too soon, the subscription is still created after the component is unmounted and left dangling in memory. I have not done thorough testing of this though.