hi all, feedback on this would be greatly appreciated: https://gist.github.com/micha/f3d9d932d8530838698a77156db0be3a

want to get bound-fn functionality in cljs there

use of js* and eval() seems sketchy, there must be a better, more efficient way to do that

@micha: there's just no way to make something like bound-fn work properly without fundamental ClojureScript support

there’s already open tickets for it but there’s no chance I will consider it until after the cljs.spec port stuff ships in a release

@dnolen thanks I'll check the tickets, but it does seem to work. which part needs cljs compiler support?

@micha it may not, but why do you need eval?

if you need eval it usually means something needs to solved in some deeper way

yeah that part is gross. i needed to resolve a js variable at runtime

anyways there’s no simple way to do it correctly without hiding all the details in ClojureScript itself if it’s going to happen

but super low priority for the reasons stated above

sure, that's why i made it a library, so i can play around with it without changing the compiler

@micha if you were just looking for feedback on your library then carry on 🙂 nothing to add!

I'm not even sure bound-fn is a win yet, depends on how it affects performance in a real application

there is a cost to using this form of binding even if you don't use bound-fn

but yeah i was hoping someone would see a cleaner way to do the eval part

@micha also I guess you don’t care about advanced optimizations?

things will get renamed and anything inside of eval is not something Closure is going to look at

(people have requested bound-fn for 5 years and I’ve yet to see a proposal that I’m particularly excited about)

it will warn if you don't declare the dynamic var as ^:export

ok so you’ve exported it

yeah it's a trade-off for simplicity

i could compile thunks in the binding macro instead

thunks that set the var

not sure how I'd restore previous bindings that way though

@dnolen is the output of cljs-in-cljs suitable for optimization in Google closure?

@micha no advanced optimizations is probably impossible

I haven’t seen anyway to make it work, the compiler needs to look up vars and namespaces

so Closure advanced optimization collapsing is out of the question

ah OK

you could off course put the output of the boostrapped compiler through advanced optimizations

but pretty sure that’s not what you’re asking

i was interested in it not so much for runtime eval as macros in cljs

yeah that's what I'm interested in

right if you want macros you have to be OK with 8-10mb of JavaScript pre gzipping

but i could run that through gcl then right?

no

I already said that above

it’s 8-10mb with Closure simple optimizations pre gzipping

(or maybe I’m forgetting and it’s 2.5-3mb?)

with simple optimizations

i mean the output of the bootstrapped compiler, without runtime eval

yes

the bootstrapped compiler is same compiler

Closure itself can run only in JavaScript, I mentioned this the other day - you could have a fully JVM-less pipeline right now granted you have time and interest to wire everything together

writing macros in cljs would be interesting because the macros could use cljs multimethods, etc

yeah that's awesome, I'll work on a boy task to do it

i mean boot

autocorrect

to be clear you can’t write macros in ClojureScript w/o bringing the entire compiler with you (not sure I communicated that)

at runtime too? or just during compilation?

you need it at runtime

compilation isn’t good enough since macros often mix runtime stuff in

surprising!

well it shouldn’t be 🙂

lisp is deep

this is why macros don’t exist in ClojureScript files!

since day 1

very interesting

i see it now

compiler macros basically guarantee no runtime dependency on the compiler

yes

man lisp is such a finely balanced thing

amazing

I'm affraid I don't understand it fully. Could you give an example?

I'm having an issue with transit-cljs. I am reading a Number value from a form (using cljs.reader/read-string) and sending it to the server which is expecting a float. If the Number is decimal (i.e. 1.12) then the server gets a float. If its is a whole number (i.e. 1 or even 1.0) then the server ends up with a long. This happens even if I cast to a float on the client. Is there a way to force transit to send a decimal type? I use transit/bigdec for money which works fine, but there isn't a transit/decimal.

hi evrybody

@cmcfarlen: there’s no way to force it

@dhucerbin: macros can depend on runtime stuff, simplest case is previously defined functions

in Clojure people do that all the time

Huh, I was thinking about it backwards. Thanks!

@dnolen: thanks. Would you recommend using a tagged value in this case to coerce to a float or normalize to bigdec?

@cmcfarlen: yes that’ll be simplest

@dnolen: yes, that was easy and works great. Thanks! Transit is good stuff.

how should i use cljs->js? i get a “use of undeclared cljs->js”..

using re-natal with reagent

clj->js

oh snap

thanks..

what libs would you recommend for animating the ellipses in the background: https://drive.google.com/file/d/0BzfCuz894WQcV2I4RHNuMjB2ZTQ/view

looking now at fabric.js

but maybe there is some simpler way...

so in re-natal, in the sample app, if i wanna set the greeting when pressing the button, i would call (dispatch [:set-greeting “bla”])?

i mean is that the way to do it. i know it works, but is that the “cljs/re-natal/reagent/all that” way?

also, i want to have persistent storage of the db, so i keep the greeting after app restart. i found AsyncStorage. is there an example of usage with re-natal?

@dimovich: CSS?