Havoc Pennington

If you make the test merge a more complicated list of keys does it still pass? I don't know, something like a,b,a,c,q,c,q,d,c,b,q If this actually deterministically associative there's no big...

Yep I think so. So what is the "rule" making that deterministic, is it that we always keep the relative order of keys in the first object, when we merge...

yeah I meant like an informal math-style proof. just thinking through whether the stated rule if correctly implemented is associative. It won't hurt to also test some weird cases of...

would it be simpler to wrap the `value` field in the constructor? though I suppose that uses memory more permanently so I don’t know.

I had no idea about this! surprising Java fact. Thanks!

Yaml libraries are just as large as this library so we don't depend on one. One approach could be to parse the yaml with a yaml lib then render it...

I think some kind of “SPI” approach where the yaml parser can be dropped on the classpath could be worth exploring

There are tons of open issues and PRs, in general I don't have time to work on them for now. Sometimes I will approve something or comment on something that's...

It seems fine to me as long while nobody's actively working on the library, someone could always put it back if they planned to do a round of work. (Side...

This may be helpful: https://github.com/sbt/serialization/blob/master/serialization/src/main/scala/sbt/serialization/pickler/JavaExtraPicklers.scala#L39