Check out the https://clojure.github.io/spec.alpha/clojure.spec.test.alpha-api.html#clojure.spec.test.alpha/instrument for instrument:

Usage: (instrument)
       (instrument sym-or-syms)
       (instrument sym-or-syms opts)

Instruments the vars named by sym-or-syms, a symbol or collection
of symbols, or all instrumentable vars if sym-or-syms is not
specified.

Cool, thank you

It’s an error:

user=>        (def arr (make-array String 8))
#'user/arr

user=>        (aset arr 0 42)
Execution error (ArrayStoreException) at jdk.internal.reflect.DirectMethodHandleAccessor/null (DirectMethodHandleAccessor.java:104).
java.lang.Long

thx

that looks like diamond problem

They do not implement the same protocols

defrecord A implements protocol A, defrecord B implements protocol B

protocol A and B have a same-named function

Are protocols A and B in the same namespace?

yes

Should protocols all be defined in separate namespaces?

I don’t believe that is possible. A protocol will end up defining a function in the same namespace. You should be getting a redefinition error

Not all protocols. But can’t have the same name in the same ns

I mentioned that I am, yes

Well there are a number of associated functions / the entire defrecord it may as well all be in a different namespace 🤷

is this your problem

this is cljs though. clj might be different.

you cannot have multiple vars with the same name in the same namespace

user=> (defprotocol P1 (f [_]))
P1
user=> (defprotocol P2 (f [_]))
Warning: protocol #'user/P2 is overwriting method f of protocol P1
P2

in the same user namespace, how would you call P1's f on an object vs P2's f?

and its giving you the answer right there: you've redefined f

user=> (ns-publics *ns*)
{P1 #'user/P1, P2 #'user/P2, f #'user/f}

it is not the location of the records

I am fully grasping the problem thanks

i guess putting all the commons of 2 protocols in another one and then subtype that will work?

there are no subtypes. you need to separate the protocol definitions into different namespaces

nor is there type composition?

well I think that depends on what you mean

one object can extend multiple protocols and/or implement multiple interfaces

composition as in “composition over inheritance”

Protocols are different from interfaces (deliberately ignoring that defprotocol creates an interface for different reasons). There is no inheritance. You cannot make a protocol extend another one. When you extend a type to satisfy a protocol, you do not subtype the protocol. In Java, classes/interfaces serve multiple purposes. One of them is namespaces. Two interfaces can have the same method name, since they are namespaced to the interface. In Clojure, these purposes are split into different concepts. Protocols provide polymorphism, but not namespacing. This is what namespaces are for. “Methods” defined by protocols are put in the current namespace. When you extend a protocol, you only register an implementation for the protocol’s dispatch mechanism. You do not “inherit” the functions, they will not be copied to the current namespace. Other than Java, you can extend two protocols that define the same function name unambiguously:

(deftype Impl []
  foo/Protocol
  (example [])

  bar/Protocol
  (example []))

(-> []
    (cond-> some-boolean? (conj {:name "Forecasted"}))
    (conj {:age 39}))

Something like that maybe?

And you don't have to start with []:

{:metrics (cond-> [{:age 39}]
                  true (conj {:name "Forecasted"}))}

@U01RL1YV4P7 - thanks for your response, will it change the order of the elements in vector I need to keep them in same order

Good catch- so use @U0P0TMEFJ's solution. Anyway, are you sure that a vector of maps with one entry is the best structure for your task?

Sure @U01RL1YV4P7 thanks again, the actual code has 200+ elements, I just gave a sample to understand the requirement 🙂

@U0P0TMEFJ - thanks much Ed

(concat (when x? [...]) [...])

that's how I normally do that 😉

it's named so you could potentially use it in expr

don't know if anyone ever actually does so

yeah i don’t understand the point of ret

seems like a totally unnecessary piece

this long pre-dates my involvement with Clojure so I am only guessing, but generally if we need a name that the caller of the macro may need to use, we let them name it (otherwise you're in anaphoric macro land). I think proxy is the only exception to this general rule (with this)