[lmm]

"s-expressions" is vaporware. Everyone who says it has a slightly different format in mind. It's easy to be all things to all people when you don't have to actually implement anything.

[thrown_22]

Every lisp program supports that serialization format out of the box.

[lmm]

No, every lisp program supports its own idiosyncratic variation on it, no two alike.

[thrown_22]

Yes, just how no two json implementations support the same schema.

[lmm]

Independent JSON implementations can parse and manipulate the same files, since they agree on what the basic datatypes are. S-expression implementations don't; often they don't even agree on what the quoting convention is.

[thrown_22]

>since they agree on what the basic datatypes are

That is not the job of the data format, that is the job of the meta data format. That this is still a draft tells you how far behind json is: https://json-schema.org/draft/2020-12/json-schema-core.html

Again, json is gods way of teaching bad devs why xml is such a mess.

[lmm]

> That is not the job of the data format, that is the job of the meta data format.

In theory maybe. In practice it's really nice that all JSON documents have the same standard handful of basic datatypes (number, string, etc.).

> Again, json is gods way of teaching bad devs why xml is such a mess.

Maybe that's the lesson bad devs take from it. Good devs take something rather different from it.