bit stumped. Why does (js/history.pushState nil nil "pizza") worked, but ((.-pushState js/history) nil nil "pizza")) throw error?

@mobileink I haven't tried, but look at this: https://github.com/anmonteiro/lumo/issues/171#issuecomment-305617325

Oh oops, I have just now read the whole conversation, ignore the above 😁

@alexwheeler is js/history.pushState a method call or a property access?

@joelsanchez Why not just use https://www.rust-lang.org/en-US/ ?

Hello all, how can I see if javascript lib for jQuery was present?

the library tells to use like this $('selector').packery() and I used like this (.packery ($ ".grid"))

but is getting ".packery is not a function"

I included it with <script src="js/asset/packery.pkgd.min.js"></script>

I used too [jayq.core :only [$ append delegate data]] for jQuery clojurescript

if anything it would be js/$

javascript globals are not implicitly in clojure scope

if jQuery is a global, you could usually also check it with js/jQuery

but if I log it (js/console.log ($ ".grid")) tell me the element

js/$?

Chrome might be aliasing js/$ to js/document.querySelector... I would explicitly check for js/jQuery

the ($ "selector") is reading

the problem is when I call (.packery ($ ".grid")) tells that .packer is not a function

*`packery`

don´t you think that the libary was not install as jQuery lib?

I discovered the problem

just page reload the lib is in jQuery 🙂

wrt jvm/node interop: Wondering if/how j2v8 might help. Any experiences?

What's the simplest way to do some generative testing of my CLJS tests in my bitbucket pipeline? I have it setup to do an uberjar build looking for compile problems at the moment. Obviously lein test works (for clj tests) but for CLJS I need to have a JS VM.

That is to say I have CLJS specs defined. I'm looking for a non-fiddly way of running stest/check from my docker script.

Perhaps this is the answer: use node. (I'd appreciate any other recommendations) https://github.com/clojure/clojurescript/wiki/Quick-Start#running-clojurescript-on-nodejs

@olivergeorge I haven't tried this, but I've seen it mentioned https://github.com/bensu/doo

some of the environments it supports are built into the vm (or are deps you can pull into the jvm)

thanks @noisesmith i'll keep that one in mind.

Quick followup. Node works if you're sure your code isn't looking for js/window or the like. That's problematic where logic which sits alongside rendering code.

My first bump was that the ns referred to a foreign lib which assumed window existed.

seems like a bug in the library?

I use lein-doo successfully to run cljs tests, recommend it

@noisesmith @joelsanchez ferret code is embedded in the emacs org-mode file - http://ferret.org

Browsing through the code right now, let's see if I can achieve "hello world"

Seems most of the ferret code is a bunch of C++ strings wrapped in s-exprs.

@deas Sounds interesting. Seems like j2v8 would allow Clojure (JVM) to manipulate node/V8 instances. Specifically, though, you wouldn't be using it from ClojureScript but from Clojure. And a lot of the reason some folks want CLJS on node is get rid of their JVM lol

What other use cases were you thinking of for j2v8?

@john Not quite sure yet. I think those targeting the browser should embrace webpack. They say j2v8 supports node modules. And js land can call into jvm as well. Might help wrt potential cljs webpack loader, figwheel, gcl and other goodies which are jvm today.

... executing on the jvm only that is.

The Nashorn based approach (Avatar) to get node modules executing in the jvm was immature and pretty much dead last time I checked.

Interesting... Let us know what you find out if you take the plunge.

@olivergeorge uncharted waters but chrome tream recently released a headless version. Guess you could use that with bensu/doo

Kanaka is one prolific wizard though. Maybe Ferret would be a good way to learn some C++

Hello all, how do I convert this to cljs ? w2ui['grid']

(? js/w2ui "grid")

(aget js/w2ui "grid") ?

hum let me see

thank you worked

@john Running a wm with Ferret, although the code is 99% C++ and 1% Clojure, but I will improve it now

So happy to see it can be done

wm?

window manager

as you can see it totally sucks, it is basically tinywm inside a string http://incise.org/tinywm.html

but hey...

Oh wow

Nice, so you should be able to decompose some of that into smaller functions and then make a more composable wm system.

Indeed, indeed

looks like fun 🙂

Build script: https://gist.github.com/JoelSanchez/9656c7096ebf9e356beef45506be4017

You can download the ferret standalone executable and execute that script to build it

Use xnest to test the wm: " Xnest -ac :2 & "

nice

you move windows with shift pressed, that's it

great functionality

hmm

I'll give it a whirl.

if I have a multmethod foo and a macro bar, what will (methods foo) be during bar's macroexpansion? In clojure, it picks up all my registered methods, in cljs it skips some

Afaik in cljs you need to explicitly require any namespaces that extend a multimethod, as opposed to clj

Not sure if that's your case

I think you need to do that in clojure as well

it seems to only pick up method registered before the macro definition

before = same file or dependencies

In clj the multimethods get extended by namespaces no namespace is :requiring (but the namespace has to be loaded, of course)

In cljs the require is needed (this is my experience)

wait how is the ns ever loaded if nobody requires it?

(clojure.tools.namespace.repl/refresh-all)

I think leiningen is able to load all the namespaces of your application on startup as well

Could never do that with cljs

I've never had lein load namespaces not in the tree of things required by my init-ns or main

and if it did, it would just be disguising errors that would happen when I deploy

Indeed I'm probably mistaken, and refresh-all is the way this gets done

Found the source of my confusion: https://stackoverflow.com/questions/39585510/using-clojure-multimethods-defined-across-multiple-namespaces This answer is incorrect as noted in the comments that I did not read

Solution works only on the repl and with refresh-all

man namespaces r weird

they are all "places"

feel like we could replace namespaces with datomicy thing where you query for all the vars you want and close over them for an ephemeral scope

and just store code and basically everything in that db

backed by in-memory or distributed or server or whatever

That's the dream

because the serialization or load mechanism for a namespace is a file

a FILE

of code

where u write the serialization format by hand

welcome to post-1960

But when the top level isn't mutable, repl driven dev would become weird and unlisply.

nah man, you write ephemeral query and transact back

then if you like what you wrote you save it

thats how you write a function or rule

;O just me saying weird things, i do that

Immutable namespaces have been discussed repeatedly over the years. Some people have even tried implementing them.

i know we tried to fly many times

eventually someone did

Indeed. I'm betting you can too!

:D yay

I think you can't change what something refers to in Haskell

Bring us home the trophy!

lolol oops feeling da pressure

@joelsanchez yea but that namespace situation is weird

datalog queries with pull over eavt seems v nice to me tbh

it gives you relational plus inheritance (recursive queries) with granular timetravel by its nature and by design

rather than by bolting on

Indeed the pull api is amazing, but i'm having a hard time imagining that as the source of namespaces

i agree tho having to query to get basic things like 'sin' or '+' would get old fast

I think in some old talk - or email - even RH speculated on the same idea... storing vars in datoms, or something like that.

the way i see it

namespaces are an affordance to bootstrap the next phase of shared data/code which would be something that affords what eavt affords, however implemented

things like versioning and dependencies (some stuff hit at in Spec-ulation RH talk) could be nicer over it

weening humans off rows and rows of text symbols is apparently non-trivial, though.

thats why u embed it into a cartoon and video game for kids

like meier's code [shifter] narrative

http://www.codeshifterbook.com

Not long ago someone posted on HN a programming language whose development environment worked by editing the AST visually, no text involved (not really related to the namespace conversation but)

(book actually uses clojure for fictional code lolol)

@joelsanchez totes

i think with cljs over react native

you can render asts from pulls automatically as draggable nested blocks on an ipad or whatever

then u r editing the shared world ast for the collective mind

kk this is where i go off the deep end 😅

Damn, found it! http://www.lamdu.org/

nice

"The main idea behind the Lamdu project is that the canonical representation of programs should not be text, but rich data structures: Abstract syntax trees."

Here's one from 1986 https://youtu.be/I9LZ6TnSP40

yeah thats nice @joelsanchez

IMO language is abt empathy and computer languages empower humans by having them empathize with a different kind of processor through their intuitions of scope and naming

Yeah, I've been arguing for editing the AST directly since the 'tunes' list.

I think one way to think about it is that visual editing for one kind of dsl is ast editing

If an element in it is a color you render a colorpicker, or a vector you render a direction picker or absolute position picker depending on metadata, etc

But ultimately in a simulation

You want to drag and drop objects

You want to paint sprites with an apple pencil or w/e

I think all of this kind of relates to http://witheve.com/

There's that. But even without inline objects, I think parinfer/paredit/other structural coding tools are just approaching the true "ast editor"

Lisp is pretty close to an ast, so lisp editing tools are a kind of "ast editor", kind of...

Yah

Yah

Anyway a more visual and interactive way to do it would be nice

eve feels a less general version of query -> transact over eavt but i think that may be too reductionist

I love that this community is willing to engage with such form of discourse lol

Like this is the sort of thing its fun to talk to humans about

I made this a bit ago on C and Lua

https://youtu.be/4LV8jJMeuRA

made a game in 48 hrs on that for Ludum Dare and it went ok -- cljs gets u 70-80% of the way there out of the box

the c lua engine i had to make the ui and gfx layer from scratch but react native gives you super solid ui really easy

lol and I don't work in a dev shop, so I almost never meet people IRL that know what anyone here would be talking about.

tbh even in dev shops you have to try consciously if you wanna engage along this dimension

Often ppl r about "current engineering practices" and incremental things

Than about the human problem

@joelsanchez in that last vid u c me editing stuff visually in the game live 😮

I work as a developer, just got a new job as a cljs developer, but I've always been working at PHP shops, no talk about ast editors, lisp, functional programming, or anything there

So yeah...

another space of prior art here is smalltalk / self

Watching the video

How's the game implemented exactly? Where does it run? Browser?

That editor looks lots of fun

How do C, Lua and cljs interact? Or is it only C and Lua?

I'm terribly confused 😛