is there a way to test if a spec exists for a namespaced key?

There is a 'registry' function

In master for the next alpha there will be 'get-spec'

Neat

I have defined an api spec for front end to communicate with backend. The messages include an auth token that I strip off soon and forward the rest of the message to internal handling functions. It would be cool to be able to create a spec on my backend that takes a spec defined for the API and removes aspects of it, like for example the auth-token key of the map. I guess the generic manipulating of a spec like a regular hash or something would be interesting.

otherwise i have to duplicate the definition of the message....trying to keep it DRY.

@fenton: why not define the spec once for the back end, then define the spec for the front end as "back end" and "auth token"?

@seancorfield: I've pulled my api spec into a third project that is *.cljc with reader conditionals so it can be included as a library for both front and backend. I don't think I want to pull in my backend as a library to my front end....

not sure if thats the best way to do it tho?

Just the spec for the shared portion is what I'm suggesting.

ahh that makes sense

So if your API and your back end traffic in a common subset of data, that spec needs to be external to both as a shared spec.

yeah, i like that...thanks for the idea!

hmmm...might run into same problem about how to make a spec like that... can't merge a spec map.

map spec.

You use s/and to specify the API must conform to common spec and also to the auth token spec. You don't merge the maps.

Alex talked about that technique here a day or two ago.

that one?

I'm on my phone so I can't dig into code.

i didn't try that approach, but wondering if that doesn't make data that looks like: [{:auth-token "abc123"} {:method :search :search-term "blah"}]

which is okay...i guess...

Since maps are open, saying some data conforms to multiple map specs should still leave you with one map?

(Replete doesn't seem to have clojure.spec or I could try that out)

hmmm, not sure, making the (s/keys :req ... might be a problem the :req part...meaning each needs all?

@seancorfield: It does, but it is under cljs.spec and slightly older at this point.

Ah, thanks @mfikes !

@fenton: if A needs key X and it needs key Y then that is additive and it needs keys X and Y - which is a merge and therefore what you want.

is (s/and (s/keys ...) (s/keys ...)) going to be special-cased to aid things like generation?

(s/conform (s/and (s/keys :req [::a]) (s/keys :req [::b])) {::a 1, ::b 2}) => {:cljs.user/a 1, :cljs.user/b 2}

(In cljs.spec in Replete)

Dang, I love that I can test that on my phone while lazing in bed on a Sunday!!!

I can't tell if that is supposed to answer my question or not :)

I can try it myself though, let's see...

@seancorfield: And if we figure it out, clojure.spec will act as an alias for cljs.spec in bootstrapped ClojureScript 🙂

It was for @fenton - I'm typing on my phone, excuse the slow responses.

@seancorfield: yeah, that looks good...

Replete is truly awesome.

what is replete?

nope, the generator doesn't work

@seancorfield: ahh googled it.

It is for lazy people who don’t want to crawl out of bed to their computer to start a REPL.

(Like me. 🙂 )

lol!

happy lazing folks, thanks for the help! 🙂

@gfredericks: the generator doesn't work for and with two keys?

seancorfield: right, which makes sense given what I know about s/and

it'd have to be special-cased to have any chance of succeeding

because the s/and generator just generates the first thing and filters based on matching the rest of them

with-gen to the rescue! 😆

I have nothing.

there'll be a helper for this in my inevitable utility library