Clojurians Log v2

@delaguardo good point about the keywords. Just merged in https://github.com/metosin/jsonista/pull/40, didn’t check if that resolves the issue

@karol.wojcik cljs not planned, but interesting idea. Interested in developing the cljs-client? could be just a new ns?

@ikitommi One of the attractive properties of transit is that it is cross-language, not just clj <-> clj. So when using tags in clj, it would be nice to be able to decode this in Go or Ruby, or whatnot. Btw, this is a real problem, not just one I made up.

@borkdude I think there could be encoders & decoders for all common types. Also, need to handle keys too, the current impl just handles values.

are there any cross-language tagging conventions out there? (besided transit, which hasn’t gained much traction outside of clojure, I believe)

but I guess in Go you could look at your own made up tags and do the custom serialization. Good question. I think JSON schema has something around content types

Would love to, but certainly not in this year. To much consumed by studies and work ;( Anyway if this implementation could be faster than transit a lot of companies could benefit from it. Especially when doing SPA SSR.

btw, the tagged-json is faster than nippy in the simple benchmarks. which sounds fishy.

https://json-schema.org/understanding-json-schema/reference/non_json_data.html

If performance is the goal, one option is to help make the current Transit faster.

great discussion, will start a Clojureverse thread out of this.

I've had this problem for a while https://github.com/cognitect/transit-clj/issues/43 which may be low hanging fruit to speed up transit

This is due to too many flushes

https://github.com/metosin/jsonista/pull/41 here is a PR to handle keyword format suggested by me

Can this be used to build a new transit server? Then you can use the transit cljs client?

https://clojureverse.org/t/tagged-json-with-jsonista-vs-transit/6946

honestly, haven't checked the transit internals for a while. Might be able to do a transit server, but would need to read the spec/source.

transit also does caching of symbols and values which compresses the payload. this is relevant for network traffic, but it may be less good for performance. so it's good take make explicit what you are optimizing for

https://github.com/cognitect/transit-format#recursive-tag-based-encoding there are two formats missing in jsonista: • string-based encoding • object-based encoding

I think transit might have ability to represent recursive data structures, which JSON lacks?