https://github.com/funcool/promesa

Hello all, is there any way to use chart javascript/clojurescript from java without having browser?

I want to use http://canvasjs.com charting to generate report, but it has to run in schedule mode, so no browser

I try to use re-frame together with reagent. In the initialization of the page I need to make two ajax calls. so, the initialization event has to return two :ajax effects - here is the code:

(reg-event-fx
  :init-db
  (fn [coef event]
    {:backend-ajax {:method :get, :url "/rest/items",:on-success [:set-items] :on-failure [:comm-err]}
     :backend-ajax  {:method :get, :url "/rest/settings", :on-success [:set-settings] :on-failure [:comm-err]}}))

@avivak First, there's a re-frame channel which would be a better place for this question Second, you might be interested in reviewing re-frame-http-fx or perhaps (more powerful, but more complicated) re-frame-async-flow Third, and now I really am getting around to answering your question, you, the writer of the effect :backend-ajax, can allow for a vector of maps to be supplied if you want. You get to design the :backend-ajax DSL any way you want, provided you then interpret it in your effect handler (I'm assuming it is your effect and not something from a library)

this re-frame-async-flow thing looks nice

@johnjelinek you can see here - https://clojureverse.org/t/any-syntax-to-match-async-await-in-creating-web-servers/825

@mikethompson Thanks for your answer! 1. I'll move to re-frame channel. 2. just for your answer. the :backend-ajax was written by me. but I wanted to keep the implementation of the effect as simple as possible (handle just one request). I thought it might be supported by re-frame. the re-frame-async-flow handle actions flow, I need them to be executed in parallel.

@avivak re-frame-http-fx handles multiple (via a list or vector of maps)

@mikethompson thanks 🙂

Are there any tools for analyzing the output of Closure to see what dependency takes what space after optimization?

@macroz: that sounds useful; if you find something, can you DM me the library name?

just asking because these kinds of tools exist for the webpack ecosystem i.e. https://github.com/webpack-contrib/webpack-bundle-analyzer

@macroz I built something recently. not much on the visual front but the table provides all the info you’d need. https://clojureverse.org/t/help-wanted-release-bundle-size-visualization/871/9?u=thheller

Looks cool! How does it work? Does it compile/zip each file separately or the whole bundle together?

it takes the optimized output file + source map and uses the source map to get fairly accurate sizes how many bytes belong to each individual source file

gzip is done for the overall file but not separately since that would not be accurate

gzipped size of a file is a fraction of the whole gzip size?

might be a bit too inaccurate but could be done yeah

I didn’t spend too much time on this yet. I only wanted to rough overview and something I can track over time.

lots of improvements could be done for sure

:thumbsup: this is the kind of tooling I meant

found someone made this for the purpose https://github.com/stephanos/minifyretracer

this sort of worked but most of the code is in unmapped https://github.com/danvk/source-map-explorer

hmm probably because it is NPM (React) modules

all cljs code seems to be nicely mapped

would need to have the tooling combine the source maps of JS deps with the CLJS maps

Hello everyone, is there a way to delete the property of a JS Object in CLJS ? Or unset! it ? In my tests I need to set the global error handler window.onerror but I want to clean up after

I'm currently using (set! (.-onerror js/window) nil) since it's null by default, but I'm wondering if there's a better way

js-delete

Thanks @macroz, didn't know it existed ! But I've just realized I don't really want to js-delete window.onerror, it should exist after my test. Not sure about the consequence of completely deleting it (messing with window properties might introduce some weird bugs down the road)

deleting it means it'll be undefined in the browser, whereas setting it to nil means it's null in the browser. May not be a problem...

maybe you would want to addEventListener and removeEventListener?

Indeed normal eventListener API... How could I forget ? Thank you !

messing with onerror manually is a bad idea in a testing framework... (`Karma` through doo)

hello everyone. happy commencement of post solstice gradual day lengthening. one of my rum :input#search-inputs changes the value of the atom :on-change with #(reset! ... ) and i'd like to also run another function at the same time as :on-change that runs a database query

is there a simple way to do this? put a watch on the atom and have thngs happen on watch? (have not done that yet)

@sova are you using reagent?

if not, add-watch is quite straightforward to use, but reagent has something called reactions that are designed for this sort of thing

@noisesmith using rum, which has > reactives

< rum/reactive but i'm not sure how it does what it do

my main question is, can i ttry to cram 2 function calls where 1 anonymous function fits nicely in my :on-change section

or does it make more sense and add more clarity to figure out how to make a separately exisiting atom watcha

can’t you make an anonymous function that does two things?

perhaps I am missing something obvious here

haha you definitely can, but i'm a newbie and question the obvious at every turn 😄

can one wrap multiple function calls with (do ) ... inside an anonymous call?

it seems like the simple thing is to just make a real anonymous function

#() is a shorthand for (fn []) and fn has an implicit do

oh hm

so #( something % ) is the same as (fn [x] (something x)) ?

=> '#(+ % %2)
(fn* [p1__47179# p2__47180#] (+ p1__47179# p2__47180#))

right

you're awesome

thanks for figuring out my question for me too haha

hey i have another questoin

while i have some clarity to phrase it right

i'm going to let people do the kind of search where results are instant

so i need to figure out a way to do it smarter than google, and locally cache results for every new word

typing in "hidden" to search you'd effectively be searching for:

h, hi, hid, hidd, hidde, hidden

later if you ask for "hid" you wouldn't need to send as detailed a list back [just changes/deltas/updates] ...

since it's still kinda brainstorming mode, do you have any thoughts on this fun puzzle?

there’s an example that dnolen put together for this, the short version is “debouncing”

wow really?! i am ecstatic to hear that stronger minds have already come across it xD

hmm debouncing

maybe I misunderstood your question actually… but usually debouncing is called for when using an API to back an auto-filled completion (sounds like maybe you need caching too)

Oh yes

Totally different question. Although that's a great idea, I actually want non bounced input

but i'll consider it more, maybe it makes no sense to cache hi and hid and hidd if all you want is hidden

this being clojure, that cache is very easy to create and manage - it can just be a hash-map with string keys in an atom

(as long as you don’t need something super fancy from it at least)

yeah i really don't need much fanciness. just some sort of append-only cache. although i am looking to datascript because the client-side db might get large enough quickly enough to warrant some more query flexibility

so is devcards basically react-storybook for cljs?