On Fri, Jan 28, 2022 at 9:05 AM Kellen Dye wrote:
> I manually unwrapped a bunch of the code while debugging and synchronizing
> on ctxt worked in that case; I haven't built new beam artifacts. I can
> submit a PR (effectively just adding a synchronized block), but am/was
> unsure if beam
I manually unwrapped a bunch of the code while debugging and synchronizing
on ctxt worked in that case; I haven't built new beam artifacts. I can
submit a PR (effectively just adding a synchronized block), but am/was
unsure if beam maintainers had some other insight into the jackson-beam
If I understand correctly, the documentation does imply that such
synchronization at [2] should not be necessary and that [4] and [5] are
correct. Of course it would be better if the docs said _how_ this guarantee
is provided, or perhaps more than that it is a matter of restating it as a
We (Spotify) are experiencing flaky tests on beam 2.35.0 as a result of
NullPointerExceptions during pipeline construction.
Stacktrace [1]. Root cause appears to be multiple threads
accessing DeserializationContext from
ValueProvider.Deserializer.deserialize [2] which, according to the javadocs