Where do clojurists usually place their fdef? - In the same namespace of their functions? - In the same namespace with “.spec” prepended? - In a separate, global /src/spec/ namespace tree? - other?

@vincent.cantin The answer is: up to you. There is no “best way” regarding this. Personally, I co-locate it with the defn.

Hello guys

how could i check the values of map only using clojure.spec, i know how can i do this for map keys names + values, but i'm in situation which i don't know what key names are but i know what types of values should be ?

e.g. for checking string values: (s/map-of any? string?)

should handle the values using s/coll-of

ah great

@borkdude thanks

is clojure.spec good alternative to type system in language like Haskell, ocaml ?

it has different applications, everything’s runtime, so you get no compile time guarantees, macros being the exception

but it’s also more flexible than a type system. different trade offs.

Also spec is still evolving, there are going to be changes, especially concerning how maps are spec’ed.

for more details about the changes: https://github.com/clojure/spec-alpha2/wiki/Differences-from-spec.alpha

would it make sense to co-develop the CLJS version of spec2 in the same repo and pull it outside CLJS itself, so CLJS never is behind on spec?

like test.check

am I right that you can’t use anonymous functions in specs anymore in spec2 so they would always have to be defined top-level? what about with-gen?

thanks for the link @victorbjelkholm429, useful information

I do not think you should be tracking spec 2 changes in cljs yet - still in flux

you can use anonymous functions

what about with-gen?

can?

?

I’m reading about “symbolic specs”, it seems that you cannot use anonymous functions in there?

you can

Symbolic specs consist only of:

Spec forms (lists/seqs), with spec op in function position, composed of other symbolic specs
Qualified keywords (names that can be looked up in the registry)
Qualified symbols (predicate function references)
Sets of constant values (an enumeration)

spec op can be fn

so those are valid spec forms

ah ok. I was confused, because of feedback you had earlier, I thought it wasn’t allowed anymore, but then I misunderstood

there’s a picture at http://insideclojure.org/2019/02/02/journal/ if that helps

it has a note about fn in the middle there

if you want to use anonymous functions at the top level with the api functions, you need to wrap them in s/spec

alright. I don’t know how much in flux things are, but if you want me to give it a try again with speculative, let me know.

(s/valid? (s/spec #(> % 5)) 10)

I’ve got at least one thing to fix still from Sean’s feedback

and we’re working through best direction for some of the things you saw last time still

so probably not worth doing anything yet

maybe using spec in an unintended way, but I ran into a situation where there were two spec forms in an s/and, the result of the first conformed the input and the second spec form in the s/and was a predicate expecting the original input, not the modified/conformed form. it feels like validation should be independent of conforming

and I don’t understand the purpose of conforming. is it intended to be like schema’s coercion? or is it to produce some intermediate form useful for spec-checking?

Re and, we have long considered having both “flowing” and nonflowing variants of and

And that might still happen in spec 2

The purpose of confirming is not coercion - it’s to tell you which choices were made during validation for optional specs or components

That is, why was it valid?

Sounds good. The flowing was definitely surprising.

Thanks, got it

A lot of design decisions