Fork me on GitHub
#cljs-dev
<
2016-04-13
>
niwinz07:04:58

It is expected behavior that nodejs target code (compiled with :advanced optimizations) with main function, does not receives the & args properly?

niwinz07:04:13

(defn main
  [& [type]]
  (cond
    (= type "init")
    (bench-init)

    (= type "knn")
    (bench-knn)

    (= type "test")
    (test-accuracity)

    :else
    (println "not implemented")))

(set! *main-cli-fn* main)

niwinz07:04:09

this code with :simple optimizations works, but with :advanced compilation the type or & args becomes nil

dnolen11:04:10

@niwinz: it should work but also why do you need :advanced with Node.js?

niwinz11:04:33

I don't need it, I just tried it and found that

dnolen11:04:54

ah yeah we don’t test that much because as far as I can tell it’s not useful

niwinz11:04:40

some peple claims that node also can benefits from advanced compilation and/or dead code elimination

dnolen11:04:51

I don’t see how

niwinz11:04:25

maybe parsing time, I don't know, in any case I don't need it and I agree that :simple is more than enough

dnolen11:04:31

I’ve yet to see anyone really demonstrate that parsing time really matters under Node.js

dnolen11:04:51

happy to look over some evidence if somebody is willing to come up with something.

nberger16:04:16

Any ETA for the next clojurescript release? I'd like to use parallel-build simple_smile. Are there blocking issues that need help?

dnolen16:04:30

@nberger: will shoot for Friday

dnolen16:04:58

I’d like to get this patch in too

dnolen16:04:01

it needs testing

Roman Liutikov20:04:40

@dnolen: looks like @mfikes successfully tested this patch with Planck and in REPL. I also didn't notice any errors when building with Leiningen.

Roman Liutikov20:04:36

Are there any more cases where it makes sense to test the patch?

maria22:04:23

@dnolen @roman01la Also tested the patch with my personal projects and things worked as expected :thumbsup: