Clojurians Log v2

I know a fair amount about perf testing on Clojure on Java, but not so much about what good recommendations might be for ClojureScript. Are there 1 to maybe up to 3 JS runtimes that are most important for perf testing? Anything like JVM JIT warm-up time to be concerned about, and if so, any recommended libraries to help do that for you?

I'm asking because of the core.rrb-vector library, which is focused on in-memory vector data structures only, so things like DOM access or GUI interactions should be a non-issue.

V8, JavaScriptCore and SpiderMonkey are the most important ones - there is some warmup time but not as slow as JVM in my experience

for microbenchmarks I'm pretty sure there's a few libs out there

I don't use anything except for simple timing + many iterations and the Chrome profiler

I think I’ve tracked this answer down before, but now can’t figure it out 100% again. What is about JS strings that makes it safe to do keyword-identical? & symbol-identical? be done on the underlying string?

ie. using identical? on the underlying string

identical? is === which strings always are compared by value

”asdf” === “asdf” is always guaranteed to be true by the JS spec

Does JS even have a standard operation that determines if two strings (or arbitrary objects) are the same object in memory?

Maybe answering my own question, but looks like === is "same object" with perhaps the only exception being for strings: https://stackoverflow.com/questions/13685079/how-to-check-if-two-vars-have-the-same-reference

numbers are also always compared by value

@andy.fingerhut not true

JS has various primitive values

for objects yes - identity

Since this discussion reminded me of it, does JS have things like Java's IdentityHashMap, i.e. a hash keyed by object identity rather than some other notion of equality?

a relatively new edition - but yes, ES6 Map

you don't have a standard JS map type that supports equality based lookup

I don’t see any concrete JS evidence that strings are identical always

but it seems to work out for the cases I’ve tried - but then again, it’ll possibly vary from runtime to runtime - so was wondering if it was a spec thing

like (keyword-identical? (symbol "a") (symbol (js/String "a")))

I tried these sorts of things - and on chrome at least it works

but was trying to find times when it doesn’t work, but if the identical? of cljs is === and that isn’t object-identity then perhaps that’s the answer

which, I do see that it is

it doesn't vary between runtimes

primitive strings are always ===, numbers, boolean, etc.

if it varied how could anything be portable?

@mikerod if you're looking for evidence you can read the JS language specification which covers this

http://www.ecma-international.org/ecma-262/6.0/ECMA-262.pdf

definitely recommend checking this if you want to know what the VM implementers are guaranteeing

@dnolen I now see that identical? is === which isn’t “object identity” for strings

so I just wasn’t thinking of it right I think, I’m more familiar with jvm and was just assuming some things

looks like the spec is clear on === behavior and strings

yep

I have read the spec some before too - just didn’t know what I was looking for here. thanks for the reference/info