anybody have any neat tricks for converting a state (`(def state (atom {}))`) to a url?

or, in general, declarative routing advice?

Hello all, why if I´m using figwheel and I load a javascript from index.html stop working after reload?

I have to reload the page again to working it back?

Do you mean javascript stop working after figwheel reload?

yes, I have a script that do ripple effect and it stop working after I save the cljs file

I tried to use breakpoint in browser, before reload it works, after reload it stops

Is there any error in the js console?

no

do I have to load it from foreing-libs?

I´m loading js lib direct from html file

If you don’t call the js lib from cljs, I don’t think you need to touch foreign-libs stuffs.

This is just my guess. I’m assuming you are using some kind of React wrapper. Is it possible that after figwheel reload, DOM is rerendered so js binding of the ripple effect is gone?

I´m using reagent, maybe js binding is gone. Is there any way to call load this .js from clojurescript?

to rebind it?

I’m not very confident about this. But may be you can use “js/…” like (js/console.log "message") (js/document.getElementById …)

Call this rebinding in an :on-jsload function that you specified in figwheel config

I´m calling it, but I think it will reload just the clojure stuffs, not the html js stuffs

Why when I save my data to sqlite does it turn a boolean to an int?

Ah, I see. Its a 'feature' of sqlite. SQLite does not have a separate Boolean storage class. Instead, Boolean values are stored as integers 0 (false) and 1 (true).

Spec is able to generate example data if I give spec as predicate vector? but it is not possible to get example data generated for predicate array?. I wanted it to generate js arrays. Any reason why it would be supported in spec ?

probably because generating mutable stuff is tricky / inefficient (requires cloning)

yes, makes sense. Thanks.

@avichalp that said you should be able to accomplish this with cljs.spec.alpha/with-gen, I think?

though you’d have to look into what is required for making an generator for arrays

@dnolen using fmap with clj->js on vector worked for me. Is it a good approach ?

@avichalp seems easy and safe to me 🙂

thanks!

@avichalp also into-array if the array is not nested

There is some hope to closed issues with cool comments? https://dev.clojure.org/jira/browse/CLJS-536?focusedCommentId=46305&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-46305

@souenzzo if you want an enhancement open a new ticket and link to the old one please - for this issue please set minor priority - thanks

@dnolen Hi, I was reading your post from yesterday about code splitting. I was wondering is there any way to do it dynamically ? I mean by just providing the entry point and compiler figure out the dependencies for that chunk ?

Also kudos on new changes.

@lxsameer what do you mean by dynamically?

the whole point is to get a production build?

@dnolen how should i provide the dependency list of a entry point ?

@lxsameer I don’t know what you mean?

you don’t need to provide it as the post says

the compiler knows

you never need to express deps - just entry points - the algorithm can assign everything else

(if this wasn’t clear I should update the post)

Ow I get it now, good this i exactly what i was looking for

if i find myself wanting a “default” implementation on a protocol, would that be a smell?

i.e. i have two types that implement a protocol with some 3 functions… but 1 or 2 of those concrete implementations are identical

@dnolen cool. this feature of clojurescript sits well with service workers . thanks man

@lxsameer a blogpost demonstrating use of this feature in combination with service workers would be super cool!

@martinklepsch I'll definitely do that

Anyone tried https://github.com/madvas/cljs-react-material-ui

It seems to work to a degree but it breaks figwheel reloading

figwheel reloads as normal, but the components don't render changes

I've also tried both React-Boostrap CLJS wrappers and they both use om-tools library, which doesn't compile (crashes with NullPointerException)

I am looking for a component library that would work for my om.next project and every single one I find is broken in some way

@roklenarcic That library works pretty well for many people. Must be some little bug on your side 🙂

It works perfectly when I hit reload

but if I add a component or change a label it doesn't show up

until I hit reload

and my compiled folder .js file has the change

before introducing this lib, I used om.dom and figwheel worked perfectly with that

Tried om.next with figwheel now on this https://github.com/madvas/cljs-react-material-ui-example reloads components well

btw I get this warning in console

Warning: Unknown prop onTouchTap on <button> tag. Remove this prop from the element. For details, see https://fb.me/react-unknown-prop

I'll checkout your repo and try

Make sure you exclude cljsjs/react and cljsjs/react-dom from your om.next

that helped with the warning

how do you run your project

anybody here use a server (i.e. ring) and a client side react app in conjunction with figwheel? When I run figwheel and my ring server they fight with eachother like my parents when I was 11

@goomba I usually run 2 separated REPLS when doing that, so then can live side by side without talking to each other like a lot of parents.

maybe I'm just being a dummy but my issue is I need to fetch content from my ring server and when I launch on fw the client app is put on the localhost:3449 and my ring server is living somewhere else

you have some options

1. you can just ignore the figwheel server in terms of content deploy, implement static page serving on your ring side, so everyone being at same host/port will just work

2. enable CORS on your server-side endpoint, so it will accept requests from anyone (or whatever you setup the CORS to accept)

* CORS is just returning a header from the page being served: Access-Control-Allow-Origin: *, modern browsers will recognize it

@madvas The reason reloading doesn't work is Compassus

I guess I'll roll my own routing

@roklenarcic You may have the same problem rolling your own

I haven’t followed the discussion but if you wanna force a toplevel reload you can probably pass a gensymed react key to your top-level element every time

It's weird how Compassus doens't affect om/dom reloading

but material-ui is affected

I’d love to get to the bottom of that issue

can you open an issue in the Compassus repo with a minimal repro?

I'll try tomorrow

my toplevel are routes of course

is there any way to have something else toplevel than routes

I have a header that is independent of routes and now I have it duplicated across all routes

@roklenarcic this is outdated but I think we managed to do reloading in this template https://github.com/compassus/oriens

Reloading worked for me too

when I had something like (dom/button nil "ABC")

now I wrap it in ui/mui-theme-provider and it stops working

and if I disable compassus it starts working again

Here;s a quick pastebin writeup

https://pastebin.com/HQKsqmCz

what's the best way to access javascript object properties?

@roklenarcic https://google.github.io/closure-library/api/goog.object.html#get

goog.object/get

(goog.object/get #js {:foo 1} "foo")

(hi matthew) 😉

why hello!

poppin’ champaign today, I’m sure (offtopic)

hey

how is that better than aget?

Well, aget is for arrays

you sure? It worked for an object for me

Yes, all the a- functions are for arrays. aget happens to work, but is is only an accident of the implementation. It is not really guaranteed to work.

By a-, I mean, amap, aclone, areduce, etc.

If you try aget on an object in Clojure, you’ll definitely see it not work there FWIW

If you’d like oget, https://github.com/binaryage/cljs-oops

Otherwise, goog.object/get is 👍

well I've been using that all wrong 😞

@pcj You are not the only one. The core libraries have it wrong as well.

I think one of the best analogies is (+ "foo" "bar") concatenates strings, but + is not for strings

So is http://cljs.info/cheatsheet/ outdated?

@pcj well that’s not an official cheatsheet, we’ve never recommending doing that

Oh okay. Thanks for clearing that up!

the problem is we didn’t disallow it and there’s a lot of code in the wild that continues to do so

to be fair, ClojureScript itself also did this and we’re in the middle of cleaning it up

and providing a flag to detect issues at compile & runtime

stupid question: why not get? i'm (slowly) working on Compleat Interop docs for cljs. should aget be deprecated for property access, in favor of goog.object/get?

Looking at the github issues for that cheatsheet it looks like the owner is aware, thanks to @mfikes.

I guess in response to mobileink's question, I would expect something like goog.object/get to be in the standard lib as oget since we already have aget

@mobileink it’s simple though not obvious

you can’t monkey patch JavaScript objects to make them ILookup

if you make get work for JS objects you damage ILookup performance cause you have to check for this special case

adding oget isn’t a good idea since what’s the point? goog.object gives you that a bunch of other stuff

none of these decisions are going to change, so further discussion is unnecessary

🙂

then I guess it's more a documentation issue than anything

thanks. not interested in changing things, just really explicit and complete doco for us simple minded folks.

@rgdelato FWIW, the docstring for aget in ClojureScript master has been updated to read “Returns the value at the index/indices. Works on JavaScript arrays.”

(Analogous to Clojure’s docstring and Java arrays.)

@mfikes oh yeah, I've never used it for objects. the docstring is good. it's just surprising to me that there isn't an equiv oget, so it takes a bit of research to figure out what the obj equivalent is

@rgdelato @mobileink is working on a new interop docs thing that we need to get around to reviewing, it will get covered soon-ish

cool cool

to be fair why get doesn’t work for JS Object is simply never going to be obvious to newcomers

its just boils down to performance

even if we tell newcomers to use goog.object/get, I don’t see how they’ll get it

sometimes you just can’t make everyone happy and this is one of those cases

in the whatever docs we write, we should probably mention the performance decision as a footnote

I suppose, even if you could write (get obj "foo") then this would become non-portable, specific to ClojureScript. OTOH (get (to-array [1 2 3]) 1) is fine in both dialects.

@dnolen re: performance. i'll have to think on it more, but i'm inclined to think it deserves a separate article in Guide. I see lots of performance related stuff here, Not essential for Ref docs, but good for Guide docs, i think.

I don’t think talking about performance is really that important to be honest not on the official site

if you want it you can get it

most people don’t care

since we made the right decisions up front for many kinds of programs

i never worry about it, so i'm good with that. i don't have any insanely great ideas for performance docs anyway, atm.

it’s better left unsaid 🙂

if you want to know the deep dark secrets you can always spend a couple of days reading the data structure implementations in the standard lib

good topic for specialized 3rd party docs. "High-Performance Clojurescript". Anybody? 😉

@dnolen re: interops docs, i've got more draft text, but i'm wondering if piling on PRs is efficient for sth still so up in the air. wouldn't it be better to do some blogs to start and turn them into a PR? thinking it would be easier for you than dealing with prs, dunno. your call.

@mobileink if you want to post your own thing that’s fine

I haven’t had time to look at those PRs because there’s a bunch of pressing release stuff in flight

new release stuff is exciting though, code splitting <3

great, thanks. here's a more general question for the community. i have some ideas about docs. they may or may not be acceptable to the maintainers. in development, i have some metadocs, involving things like a list of stuff to cover, a general strategy/plan, etc., which would not be included in the final result. also various drafts which may change drastically. my question: what's the best web service for this kind of thing? assuming it would become a community effort. just using github or a blog seems inefficient to me. i guess i'm asking about a good collaborative doco dev service.

seems to me like collaborating with an existing site like clojuredocs or something would be a good idea, since they already have the SEO

I’ll reiterate @rgdelato’s point - going off an doing a new thing is fine - but probably won’t have much impact

submitting something to the official site will have a lot of impact - you might not be able to write it exactly how you want - but that’s just a tradeoff

But I specified "selfbot-cljs.core" as my :main namespace, unless it's talking about something else :thinking_face:

Oh, maybe I should have asked in #leiningen

is there a compiler option that allows me to run advanced compilation but without mangling the variable names? I want to make a compilation of some Node.js code in advanced more for the performance optimizations, but I don't mind at all about reducing variable names (since in a Node.js the size doesn't matter so much)

@wilkerlucio there’s a debug flag that doesnt fully mangle names: psuedo-names https://github.com/clojure/clojurescript/wiki/Compiler-Options#pseudo-names

@lwhorton thanks, I'll try that 🙂

is there a "get-in"-style function for JS objects, or would you just some-> a bunch of goog.object/get calls to safely get a deep property?

https://lodash.com/docs/4.17.4#get ?

https://google.github.io/closure-library/api/goog.object.html#getValueByKeys

it’s probably worth bookmarking this page 🙂

awesome, thank you. I was on that page, but for some reason, I didn't see it (Note: goog.object/getValueByKeys seems to throw on a nil object while goog.object/get returns nil)