Andy Chambers
Andy Chambers
Bumping the confluent version has traditionally been a fairly simple thing to do. Here's the PR for the last time it was done. https://github.com/FundingCircle/jackdaw/pull/217 It's worth reading the upstream changelog...
Hey Mark, Looks like you've been digging into this quite a lot :-) Thanks for your interest. I think there is an arity of the serde constructor which accepts a...
I think changing this behaviour would risk breaking existing programs but it shouldn't be difficult to get the behaviour you'd like using something which strips out the nils after deserialization....
> Spec prefers you to remove the nils. Remember that spec also has the `s/nilable` constructor which might be a better match for avro's `["null", "some-type"]` > Would you entertain...
I think having lots of different namespaces all with function of the same name (in this case `serde`) is a bit of an anti-pattern in Clojure and if we're going...
I had a shot at doing this but ran into trouble implementing it in a way that keeps the tests working (and therefore breaking existing uses). I'm inclined to close...
LGTM. The existing integration tests should cover this change. e.g. this one which uses the `kafka-streams` function through the test fixture https://github.com/FundingCircle/jackdaw/blob/master/test/jackdaw/test/transports/kafka_test.clj#L54
Hey Sam. Sorry for the radio silence so far. The other person in our company who might have an opinion on this is away on holiday at the moment. I...
Good summary of the issue. Thanks @kidpollo Can we make sure the functions proposed above maintain the existing behavior if called without a `join-name`?