@xiongtx if you s/exercise it will complain about "no spec for key".

It is design decision, so you could opt in to writing detailed spec, instead of being forced to.

@tclamb seems like a bug, but would be curious to see the actual value causing the error

@alexmiller s/exercise is generating maps with clojure.lang.LazySeq keys, like {(lazy-seq '(:r)) :r}

that passes s/valid? and s/conform

I didn’t expect this to be true as well: (identical? (s/conform spec x) x)

but I think that’s an artifact of the map value spec being keyword? instead of e.g. s/cat

I ran into this speccing a map-of point to tile for the board in a game like scrabble

hey all I am trying to wirite a spec that would satisfy some json data in which the keys have a shape but are not defined eg

{
        "x-identity": "2"
        "x-name": " aaa"
        "x-[some other string]": "bbb",
}

Can I use something like

(s/def ::x-object (s/keys ???)

%(string/starts-with? % "x-")

I though about using s/cat but that implies order of the keys

how about (s/map-of (s/and string? #(re-matches #"x-.+" %)) string?)?

@tclamb yeah just figured that out

thanks

now what if I have a json with a mix of well know keys and patterned keys like the ones above eg

{ 
         "name": .. , 
        "age":...,
        "x-internal"....
        "x-created"... 
}

I am modeling the well-know part of the spec with s/keys and the other with s/map-of however s/merge doesn’t seem to be working given that the “optional x...” could be 0, does merge understand optional?

the docs sound like s/merge only works with s/keys, not s/map-of

https://clojure.org/guides/spec#_entity_maps

hmm this is harder than I though

I think I have to drop-down to everkv