is it still the case that core.async go blocks have the potential to leak memory and so should be used sparingly? Id really like to be able to just map a set of functions such that they all pass to each other over channel interface, but if sparing use of async functions is recommended, then I would optimize along putting more transformations into a single function

better question. once you put channels on channels WHY is it so hard to get things back to a channel of values??

I think the memory thing was fixed in one of the last rounds of updates?

outstanding. time to abuse channels

Hi! Is this expected? nil as a value is "null", but as a key is "nil"

(->> {:nvalue nil, nil "nkey"}
     clj->js)
;; prn => #js {:nvalue nil, :nil "nkey"}
;; console.log => {nvalue: null, nil: "nkey"}

I wonder, is there a good cljs alternative to Gatsby (data fetch & transform > graphql > static React site)? Thanks!

any transit pros know if I’d doing something wrong with custom tagged values? I’m thinking i stumbled on a bug with the :json reader https://gist.github.com/khardenstine/557fcaa221f51c48d2b1ff6bbfc102cb

How to do it in cljs?

class Foo {
  static bar = {x: 42}
}
myFoo = new Foo()
external_call(myFoo)

(deftype Foo []) (goog.object/set Foo "bar" #js {:x 42}) should work

this is a bit tricky with closure :advanced and may need tweaks

Ok. the lib will just myFoo.bar and expect to receive {x: 42}

hmm that may not actually work then. thought it was using the class Foo directly not just the instance

new thing for npm packages that claims to bring some clarity to the mess https://github.com/pikapkg/web <- @thheller

hmm interesting. don't think that its actually a good idea though.

in general, or from a cljs perspective?

the "Performance" section is a bit naive

seems to be centered around optimizing HTTP/2

yeah but it still discounts DCE completely

Uses rollup under the hood

yeah but it seems to only use it for npm deps. not project code. so it can't possibly tell what is going to be used and what not.

so "big" packages like antd or @material-ui/core will just still import everything I guess?

also there a plenty of packages that don't have a module entry in package.json. eg. react

It might pull in deps lazily, ala Rollup's "Live Code Inclusion"?

LCI being an alternative strategy to DCE

https://github.com/rollup/rollup/issues/50#issuecomment-123467947

Interesting how they mention that Rollup was partly inspired by @dnolen

I guess GCC's version of LCI is the module loader system. We just don't used it enough...

:modules are definitely not used enough yes

https://github.com/swannodette/swannodette.github.com/blob/master/_posts/2015-01-06-the-false-promise-of-javascript-microlibs.md couldn’t find a published link

http://swannodette.github.io/2015/01/06/the-false-promise-of-javascript-microlibs

Hah thanks, I guess it’s not listed in the archive for whatever reason.

Listed under a different title: January 6, 2015 » JavaScript Modularity Shaming

Damn you’re good!

well the hint was in the .md you just posted 😉

> title: "JavaScript Modularity Shaming"

that is a jekyll setup, I use a similar setup in my blog so it is familiar 😉

Oh cool, can you share the link? I used perun for mine, not super familiar with jekyll though I see the issue. The link from the github repo has a trailing slash which was causing the not-found page. 😈 in the details