Gary Burgess

There's a PR to add it https://github.com/garyb/purescript-codec-argonaut/pull/31/files but it's unmerged... I did want to do something to make it more convenient - so that codecs can still be defined record-style...

Slack must have eaten my reply 🙂 - the relevant section is here: https://gist.github.com/garyb/55d24a0a81505bf88f4e9b22f4a6b525

I'm not sure (3) is necessarily a problem - if you're telling people they have to kill forks / unsubscribe from event source in a finalizer, it's harmless, just unnecessary.

(2) is a little more subtle than you outlined too - the deadlock is one possible problem there, but only with reciprocal initialization dependencies. The more common problem is that...

It would mean introducing something similar to the `Left` / `Right` for `variantMatch` distinguishing between nullary and unary constructors, although I would probably introduce a ADT with better named constructors...

In this scenario you need to write a "custom" codec - by which I mean one that isn't built entirely out of library-provided primitives, and accordingly you become responsible for...

I don't really want to add a dependency on that here personally, but yeah, that would be an option. It'd still suffer from the possibility of overlapping values depending on...

`addDefaultField` doesn't operate on codecs for object/records (they work on the basic "any" `Json` type) so we can't do that with them in their current form. Adding a version that...

I'd say "it depends" - which is why this library doesn't provide codecs for the datetime library. If you're dealing with an external API that representation might not be suitable....

I don't know, it's just what I happened to come up with on the day I wrote it.