Your feeling on there being no way to transport this over the wire is puzzling to me but I admit I don't have all the details. My feeling is why not? Surely if we have wire formats for self describing binary objects that can then be serialized into an in memory structure, transporting a spec shouldn't be harder than that?
Not that there's _no_ way to transport it over the wire, it just requires the full environment (a JVM on the other end - because we have java-specific stuff like spark calls). I'd put it at an order-of-magnitude more complex than something declarative like a JSON schema which is pretty safe to execute anywhere.
I don't think this is a really big failing of spec - I don't know of ANY validation tools that don't have to compromise between power/extensibility/ease-and-safety-of-execution-somewhere-else.
Maybe if you implemented some kind of uber-validator in purely functional prolog or something?
Again, apologies if this sounds ignorant, but we have pretty standard practices now for sandbox execution of untrusted code. A LISP seems especially suitable for this type of task.
I don't have a clue how you would implement this. The difference in portability between spec and something like json-schema/protobuf/avro is that you can serialize the schema in these and then clojure and (say) python, go, java, C#, JavaScript applications can talk to one an other.
How would you propose to serialize clojure spec's and use them from a python app? Port the clojure compiler to python?
I second this emotion -
"How would you check a spec from anything other than clj/cljs" is (IMO) the critical question here. Sure you could check out my git repo in a safe VM and execute it there, but that's a WHOLE lot more hassle than an XML or JSON schema. It's not just a language barrier thing.
There's nothing stopping spec predicates from making network calls, looping forever, etc. If I wanted to be able to call my spec from other apps I'm writing, I could package it as a library easily, but a workflow like rest call->get spec->validate data (which I've implemented many times for JSON schema for simpler things) wouldn't really be practical with spec (without at least setting some really tight restrictions on what features of spec you're allowed to use)
Again, not really a failing of spec, it's just not designed for that kind of workflow.
SCI (Small Clojure Interpreter) is 4500LOC and if you just want a barebones clojure interpreter to carefully evaluate '(> x y), you could probably fit it in 100 LOC clj. Ok you want to use Python, so 500 LOC py. Or port SCI, 4500LOC port is a few person-months given a reference implementation.
