Fork me on GitHub
#off-topic
<
2018-04-12
>
gklijs03:04:32

A lot of especially small government intuitions are pretty afraid of GDPR, a lot of companies make some profit of it. But a lot of companies even in Europe are not fully compliant, and some big ones aren't even started to really make an effort to. One of the nice things about GDPR is that customers should not only be able to see witch data a company has on them, but also kan ask to either delete it, or move it to another company. Also companies should be able to tell which personnel information ends in witch systems, and especially what might go to other companies.

andy.fingerhut05:04:55

Weird moment for me: My wife just learned the word "concatenate" tonight - she has been a school teacher for writing, grammar, spelling, literature, etc. for 20 years. My college age sons don't recall coming across it before, either. I feel like it is as common as air, likely because of my programming background.

andy.fingerhut05:04:40

She is learning Python, which is why she came across the word now.

😁 12
vemv06:04:14

And did she know about conjoin? 😅

andy.fingerhut07:04:48

Not that one, either, but I suspect it isn't nearly as common among programmers as concatenate, either.

gklijs07:04:33

Any recent news on clojure(script) in relation to wasm? I could not find any easy way, in the meantime Kotlin seems to have a nice way to go to wasm. I want to play around a little and would slightly prefer to to it with Clojure.

tbaldridge12:04:30

@gklijs it's interesting to think about, but wasm offers little (if anything at all) to ClojureScript. Lisps (at least the fast ones) rely heavily on optimizing JITs and GCs, you give both of those up with wasm

gklijs12:04:56

@tbaldridge GC is coming to wasm, I don’t have any use case, but just want to play around with it.

tbaldridge12:04:41

having a GC would be nice, but without a good JIT we're still in the situation where there's little reason to port. Clojure functions tend to be small and very generic, these functions lend themselves quite well to inlining found in JS JITs.

tbaldridge12:04:39

So to answer your original question, no I'm not aware of any CLJS on wasm attempts, mostly because there's little benefit in performing the port, since the end result is access to the same (or really fewer) platforms that we have access to today.

tbaldridge12:04:20

That and I think few people have the desire to write their own GC and optimizing compiler.

3Jane12:04:06

> few people have the desire to write their own GC and optimizing compiler. the mind boggles

hubert13:04:03

@andy.fingerhut so, the next word will be recursion 😉

john13:04:45

@tbaldridge are there any workloads you could imagine sending off to WASM from CLJS? Some super hot loop or something? Low latency?

tbaldridge13:04:08

sure, probably the same things you'd send to C on the JVM. Stuff were you need fine-grained access to memory, have statically typed code that needs to run fast, etc.

john13:04:51

Right, and for that, you'd probably usually just compile the other thing and call it directly via postMessage from CLJS.

gklijs14:04:44

Yes, and the kotlin compiler already gives you some js to call the wasm. I’m thinking I may do something with a stic typed game state, and use wasm to calculate the next state, could be something witch mimics gravity or something, or some ai. In that case I could use kotlin for the wasm part, and use cljs for the rest.

john14:04:46

that's pretty interesting

john14:04:57

I think a jvm on wasm is going to be inevitable

tbaldridge15:04:43

I really doubt it will be very fast if they do. Remember, the JVM leans heavily on the JIT. Any mostly dynamic language (and the JVM is one of those) is going to need a really good JIT to get good performance on wasm, not to mention stuff like code patching.

tbaldridge15:04:53

All those things take years to develop.

tbaldridge15:04:24

unless you port some existing tech via something like ecripten (sp?)

john15:04:10

non-trivial, for sure

john15:04:26

the question is whether it is a valuable target for Oracle

tbaldridge15:04:58

at this point wasm is pretty slow. I don't know that I've seen any benchmarks that show wasm being faster than properly optimized js

tbaldridge15:04:22

at least for non-trivial work

tbaldridge15:04:26

Sounds great so if a restricted subset of JS with no GC or JIT runs 20% slower than a JS engine running a dynamic language. Why would I use wasm?

tbaldridge15:04:58

That's my biggest problem with the tech, it's cool for game companies who just want to get something working in the browser, but I haven't seen a compelling reason to switch normal code yet.

tbaldridge15:04:26

Maybe if they introduce threading or something. 20% perf penalty to get true shared memory threading would be pretty nice

john15:04:57

20% slower than native

tbaldridge15:04:45

ah, good catch

john15:04:53

but yeah, typed array work in js can be pretty much nativish

john15:04:08

I'm sure there are examples where js just blazes

john15:04:59

that's what makes asm.js possible

john15:04:27

in fact yeah I think there are some examples where asm.js is still preferable over wasm, for the time being

john13:04:17

Maybe some audio processing

andy.fingerhut13:04:48

@hubert I will probably ask her to tell me when she comes across a word for the first time now, to see what other words fall into this category of "my software development colleagues and I say it frequently, but many other people never encounter it" (not acronyms, which I expect, but English words that have existed for a long time like "concatenate"). Oh, and of course there is the fun quote: "In order to understand recursion, you must first understand recursion"

flowthing07:04:02

I wonder if "orthogonal" is one of those words... I'm an English major and I don't think I'd ever even seen that word before I started my programming career. It's pretty popular in software development, though, in my experience.

flowthing07:04:40

Maybe different for a native speaker, though.

andy.fingerhut13:04:51

I also discovered the Google Ngram viewer, which is cool. Here is the frequency of the word "concatenate" in Google books since 1800: https://books.google.com/ngrams/graph?content=concatenate&amp;year_start=1800&amp;year_end=2000&amp;corpus=15&amp;smoothing=3&amp;share=&amp;direct_url=t1%3B%2Cconcatenate%3B%2Cc0

hubert13:04:19

heh, I was about to write that quote

3Jane13:04:24

yeah… depending on your specific niche, you learn interesting vocabulary

3Jane13:04:04

When learning English, I skewed towards historical texts and early fiction (I like gothic horror, don’t judge me). The result is I knew what a thurible was, but had no clue that the first school you go to is called “elementary” (negative points on a test, the injustice still hurts many years later XD )

andy.fingerhut13:04:12

The Google Ngram chart for the word "complect" suggests that Rich Hickey traveled back in time to the 1930s and wrote lots of books 🙂 https://books.google.com/ngrams/graph?content=complect&amp;year_start=1800&amp;year_end=2018&amp;corpus=15&amp;smoothing=3&amp;share=&amp;direct_url=t1%3B%2Ccomplect%3B%2Cc0