note macro arguments don't get evaluated

play around with macroexpand-1 to see what your macro gets expanded into in the two cases

so how do I get them evaluated?

evaluation happens at runtime, you can evaluate arguments with a function call, for example

otherwise you'd need eval, but you're in cljs

it looks like you're trying to generate code at macroexpansion time for a runtime value

I can't see how that could work

oh… I see

that’d be sad if I can’t do it

let me take a step back

are you trying to reduce the boilerplate necessary to generate material-ui bindings?

yes

I wonder if it's good enough to do codegen the old fashioned way. discover your MaterialUI keys in an earlier build step, then serialize it to a file such that you are invoking your macro with a literal argument

¯\(ツ)/¯

why can’t I somehow loop through (js/Object.keys js/MaterialUI) in a macro?

you can, if you know the value of js/MaterialUI at macroexpansion time

getting the value through an argument to the macro delays its evaluation so you don't know it

so I have to do it in the macro

I’m trying to figure out how to loop in a macro… I looked how doseq is done… and… that’s some scary shit

here's an example derived from your post

https://gist.github.com/arbscht/1a1fb5fa99080900ba965b961f8c4f6f

wow, thanks… I’ll study it

(tags is redundant, try-me doesn't use it)

right

that doesn’t compile though.. it doesn’t know what js is

right, so it won't be able to find MaterialUI either in that case

(I'm testing in a repl which has js loaded but the compiler won't, of course)

if you used a clj hash map rather than a #js array, it should work

I give up… I’m tired and I can’t solve it… even tried using js/eval. I’m missing something here. Will try again later. Thank you for your help @UHBQQMPGR

sorry I can't be of more help this time, also a bit distracted myself. the macro-eval-compile triad in cljs is a tricky beast

You helped me It is tricky indeed, but I intend to solve this puzzle.

fwiw, I'm using material in a project too, and have gone with the codegen route I suggested earlier

it's not very neat though, I can see the temptation to do it dynamically in one pass

It might turn to be unsolvable. Whatever I come up with if it doesn't pass advanced compilation is useless anyway

https://dev.clojure.org/jira/browse/ASYNC-74 unfortunately no

Thanks