is there a known time scale (or terminology) that makes a single millisecond sound expensive? e.g. 1ms = 1000 blubs My function takes 20000 blubs to run -> sounds nasty, will fix Conversely, 0.089405 msecs is not very expressive, it makes me perform a mental multiplication in order to figure out if the fn was expensive (in "computer" timescale, not in human timescale) or not

@vemv clock cycles?

😄

(defmacro clock-cycles-of [& forms] ...) easy right 😜

easy peasy

you could also translate the running time into something physical

😄

e.g. “your function took as long to run as….”

@vemv What's wrong with microseconds (µs) or even nanoseconds (ns) 🙂 ?

@me1740 microseconds sound great! had forgotten about them, I see them only occasionally in programming nanoseconds would be still confusing... e.g. (time 2) -> 150000 nanoseconds, too big of a number to constitute a baseline

The same metric prefixes apply to all standard units so it's worth knowing: https://en.wikipedia.org/wiki/Metric_prefix

I played a bit with Criterium. Yes one can quite easily get used to deal with nanos and micros (which Criterium uses interchangeably) Wasn't that satisfied though... so I "came up" with this (apostrophes because it might be common practice)

(defn throughput [microseconds]
  (-> 1.0 (/ microseconds) (* 1000000)))

Is it reasonable to assume that you'll never find EDN where a key in a map is another map?

no

What's a use case for this?

you can have arbitrary data as keys in EDN, so it would be very brittle to assume this would never happen

clojure.set/index produces indices with maps as keys, which I may want to serialize for whatever reason

unless you specify a schema that you enforce yourself between (de)serialization, but in general you should not make any assumptions

(edn being a serialization format for clojure's datastructures, which everyone wants to use as the name for clojure datastructures for some reason)

In this case, I mean EDN-EDN, as in https://github.com/edn-format/edn

As it turns out, it's not always easy to model EDN in other languages.

there have been a few other impls: https://github.com/edn-format/edn/wiki/Implementations

Does anyone have any experience running gorillalabs/sparkling on an EMR?

I am fairly new to spark, so I am not sure what the best way to put a repl on the EMR is

(the best -> any)

Has anybody ever used https://fission.io for functions in docker containers/kubernetes?

What is that talk where Rich Hickey says something along the lines of "the person with the simpler system will drastically outperform the person with a more complex system"?

(I mean, that's the whole theme of "Simple Made Easy", and most of his talks, but there was a particular quote about "outperforming" that I'm looking for)

maybe grep here https://github.com/matthiasn/talk-transcripts/tree/master/Hickey_Rich

I love lodash but...this is their cond api? https://lodash.com/docs/4.17.11#cond

I guess clojure's cond is a macro, so you have to have such a funny api to do something similar in js

conforms is kinda awesome

it's got pattern matching built in. i kinda like that