How to get into graphs in clojurescript? I'm using ubergraph to describe some control structures of my system and in the moment, I plot from repl with uber/viz-graph. I want to put it in a "admin pannel" in clojurescript. But I didn't find: - how to/what format use to export graph (with metadata) from back to JSON/EDN/transit - how to plot it in clojurescript. There is some stuff on cljsjs but each lib has it own input format for graphs...

@souenzzo depends a lot on your needs. If you're happy with something generic looking and aren't sweating the details, then just hit up cjsjs and search for "chart"

if you want to make something more bespoke, I've had an OK time wrapping the lower level utilities of d3 to calculate things like scaling x/y positions and then rolling my own SVG elements with attributes

or, to put it another way, there is no "standard chart library" that works for everyone in javascript, so there isn't one in cljs either, so there is no "standard chart input format" either 😞 - although there certainly are some efforts to do so, such as d3 and https://vega.github.io/

Clojurescript on Windows users. Please vote on this. https://dev.clojure.org/jira/browse/CLJS-2036

It looks like a compile bug associated with the :preloads feature. Stops us from using the more recent stable releases of CLJS.

@olivergeorge we have a very small number of contributing devs on Windows machines, happy to see that one fixed but you’ll probably have to provide a patch yourself. Pretty sure this is likely a file system issue, that’s been most of our problems on Windows.

Pretty set on clojurescript at the moment, but does anyone here have any experience/views on ScalaJS?

Hi, can someone give me advice as to the best way to give a sizable financial contribution towards clojurescript development? I don't see any info on this in the github page or the http://clojurescript.org site.

I believe @dnolen is the lead dev. He could probably point you in the right direction.

@qqq :thumbsup: I'm reachable at <mailto:[email protected]|[email protected]> if @dnolen or anyone else wants to get in touch

@drcode : if you are throwing $$$ at clj/cljs, another project worth looking at is: https://github.com/clojure-emacs/cider ; they have detailed "funding" page at the bottom [disclaimer: I use cider; I am not a dev of cider. I would benefit indirectly (via improved dev tools) but not directly (i.e. cider does not pay me.) ]

@qqq I also use cider every day- I will keep that in mind as I plan things out.

I'd also advocate throwing money at figwheel. Not sure if Bruce is soliciting, but he's providing an important service to the community IMO

@john yes, already contacted bruce 🙂

What's the simplest way to turn CLJS into JS? I'm trying to reproduce a bug, and I'd like to avoid a whole project setup if possible.

@peeja that's one of the first examples in the clojurescript quickstart https://github.com/clojure/clojurescript/wiki/Quick-Start

just requires cljs.jar

@noisesmith Ha! I think I clicked every link on the wiki but that one. Thanks!

this is also easy: http://app.klipse.tech/

It looks like when you take the value of a dotted interop form, say .aJSMethod, ClojureScript emits simply .aJSMethod

So (println .aJSMethod) emits cljs.core.println.call(null,.aJSMethod), which is invalid JS

Is that a known issue?

Hmm, I'd expect to see quotes around .aJSMethod at least.

I'm not sure what to make of this: when I put it into http://app.klipse.tech/, I get an analysis error, but also output.

.aJSMethod alone gives the error and no output

Unexpected token .

the reader chokes on it

printing a dot method, unassociated with any object or anything, I don't think would be expected to work.

or, well, no the reader doesn't choke on it

javascript's parser chokes on it

Added here: https://dev.clojure.org/jira/browse/CLJS-2070

Right, it shouldn't do anything useful, but it should break the build process correctly.

the answer I expect to see there is that .aJsMethod isn't a valid form

garbage in, garbage out

Yeah, exactly

Well, no, hopefully garbage in, error out 🙂

Ideally, never garbage out

clojure and clojurescript do a lot of silly things instead of outright rejecting inputs

in a lot of contexts

spec may change this in the long run...

But there are probably a billion ways to cause a js parser error. Do we catch all of them?

Why would the compiler ever output bad JS?

except in the case of a bug

I doubt we catch any of them - explicitly at least

It's not compiling, with println

?

How is it not compiling?

(println .something) is not "emitting" js

No, no, that form is compiled, and the compiler emits bad JS

oh

println could be anything

see, this is why using cljs.jar make sense - you can say that with a specific version of clojurescript we get this specific output...

oh, how'd you get there?

@pedroteixeira did you ever solve your issue with running lein cljsbuild once hello once you cloned the om repo. I am still not able to figure out this clojure.lang.ExceptionInfo: Invalid :refer, macro cljs.pprint/deftype does not exist in file examples/hello/out/cljs/pprint.cljs {:tag :cljs/analysis-error} exception.

For context, here's the bug surfacing in another issue, which is what led me down this rabbit hole: https://github.com/funcool/promesa/issues/48

Now, granted, Promesa's macro acrobatics here may not be completely thought out, but it shouldn't be able to emit invalid JS

$ cat src/repro/repro.cljs 
(ns repro.repro)

(println .fooBar)
$ rlwrap java -jar cljs.jar 
Clojure 1.8.0
+user=> (require 'cljs.build.api)
nil
+user=> (cljs.build.api/build "src" {:output-to "out/main.js"})
nil
+user=> 
$ cat out/repro/repro.js
// Compiled by ClojureScript 1.9.562 {}
goog.provide('repro.repro');
goog.require('cljs.core');
cljs.core.println.call(null,.fooBar);

@peeja I'm going to take a wager that the cljs devs disagree with you, and that it's handy to reject invalid input, but there is no priority in cljs or clj dev to make erroneous output impossible

Seriously?

I could be wrong though, but I've seen similar responses to various other clojure / clojurescript behaviors when given erroneous input

I would be extremely disappointed to hear that that was the case.

"it does something invalid for this input" "that input wasn't valid so it's not a priority to fix it"

it's a conversation I've had more than once with clojure devs

Explicitly de-prioritizing error messages for problems that are obvious to the complier and non-obvious to the user is a great way to tank a language.

Is this valid input: (.aJSMethod (js-obj))

but is it obvious to the compiler?

does cljs have the infrastructure to recognize valid vs. invalid js forms?

but also you can impl it yourself if it's a lisp right

Sure it's obvious. It does something special with (.aJSMethod (js-obj)). It doesn't do that with (foo .aJSMethod). Somewhere in the compiler there's already logic that says what to emit for that form, and that logic is incorrect.

if there is a finite amt of time, it sounds more like they prioritize other things more, and the explicit deprioritization -> tanking argument may be a lot

Why does it emit .aJSMethod?

i generally give them benefit of doubt

The analyzer already knows that that symbol can't be evaluated there. So why does it emit anything?

how does it know?

it's an interop

does the analyzer run after full macroexpand only

I bet it's related to the fact that (js/console.log ...) works

It's not a valid interop form

which is totally invalid as a clojure form

but works (accidentally?)

How does it know it's not a valid interop form?

@peeja cljs doesn't assume a closed world where it knows what exists and what things mean in the output code

it happily outputs things it knows nothing about

atomically, (after macroexpand) there are a finite number or patterns to match interop forms against and it's not one of them @john i believe

they are probably implemented by some other lib somewhere...

CLJS just passes things through that it doesn't recognize?

That doesn't sound right

all the time, yeah

Like what?

debugger doesn't work

usage of external libs

There's an explicit reader form to emit raw JS

The JS side handles a lot, obviously. Just like Java in Clojure.

Otherwise, the analyzer expects to understand it

It's a very host-dependent language.

@peeja sure - but the compiler happily passes things through as is. I'm not objecting to the very idea of validating the compiler's output, but cljs has no such infrastructure or functionality right now.

I'm not talking about what should be, but what is there.

"but the compiler happily passes things through as is" -> Can you give an example?

it's got near zero validation in place

I'm not talking about validation at all

@peeja as I said, pretty much any usage of a lib that isn't part of js

it just passes through all your references

Right, but symbols which begin with . are treated differently in CLJS

Already, today

only sometimes

only when they are the operator right

:thinking_face:

eg. js/consolre.log - the clojure language says that shouldn't work, but because it's passed through it works

Also, hang on, random symbols aren't passed through to JS at all

That's just incorrect

but we have evidence that they are

Do we? What's that?

I've never seen reason to beleive they aren't

Have you ever seen anything else pass through directly to JS?

I can emit calls to libraries that cljs.jar knows nothing about

How?

by putting them in the code

Using js/, you mean?

@nikki can you elaborate? re: a finite number of patterns?

You can't just say Date and have that pass through to JS

yes, that's true

type .aJSmethod in the repl and it'll pass right though to the JS site, with an error coming from JS land.

@john https://cljs.github.io/api/syntax/dot (x .y) doesn't match any of those patterns

Yes, that is the one example we've got so far 🙂

Because any dot prefixed form is potentially unknown to CLJS

But dot-prefixed symbols aren't passed through to JS in any other situations

only when they're illegal JS

?

oh, otherwise they get converted, right

i think it whitelists interop forms and does the interop, otherwise passthru

@nikki I believe @peeja was saying that a (.x y) form was causing the erroneous output, when .x doesn't exist.

When the compiler compiles (.aJSMethod something), it does become something.aJSMethod(), but the compiler's not just blindly passing the characters .aJSMethod through between something and (). It understands the JS it's constructing there.

i thought that the macro changes it to the other form before js emit

@nikki What I'm trying to say is that there's no other examples of passthrough, so I don't understand why that appears to be the fallback

+user=> (macroexpand '(println .fooBar))
(println .fooBar)

:thinking_face: yeah i meant that as an accident of the current constructions

If that were correct, you'd expect to see passthrough happen in other cases, producing correct output

some times the way things happen isn't understandable with meaningful reason... lol

@peeja whats the most macroexpand'd version of the bug example with p/alet?

Which is why I think there must be a branch of the logic which already addresses this case, the only case where passthrough happens, which could throw an error, but instead says "ah, screw it" and emits the bad JS instead

@peeja what do you mean "It understands the JS it's constructing there."? In what way?

if u do (println x) doesn't it pass x thru or am i wrong

r u saying it should know that .x is a bad form of JS identifier

It doesn't, the analyzer figures out what x means. It could be a local or a var.

if it's not a var, what happens

If there's no local, it assumes it's a var

ah its the other way round? ok

Unless I'm wrong… checks

cljs.user=> foo
WARNING: Use of undeclared Var cljs.user/foo at line 1
nil

Yep, it looks for a var (and in CLJS, a missing var will evaluate to nil)

mmm

we r guessing that based on an example

So I'd understand more if it was looking for a var called .aJSMethod in the current namespace.

But it's not, it just emits that name literally into the JS. Which is weird.

Have looked into the guts of the analyzer and checked out what it is attempting to do with dot prefixed forms?

(thanks @john for letting me know about app.klipse lol this is cool)

@john Not yet

hmm

on another topic: other than http://thi.ng and kovasb gamma, do y'all know some neat paradigms / libs for webgl on cljs?

http://regl.party is data-oriented enough that it can be used from JS interop from cljs pretty well

@peeja might want to bring it up in #cljs-dev. Like we've been saying, CLJS doesn't have a general policy of cleaning all garbage on the way in, which, as @noisesmith mentioned, may be improved on with spec in the future, but it couldn't hurt to ask them.

spec plz

Right, I understand that, my assertion is that I don't think this actually needs to be cleaned on the way in so much as handled differently where it's already handled. But I'll dig deeper.

Frankly, I don't see the issue. The error is one level later. At compile time of the JS code. If this was a runtime error... I'd understand.

where is the most thorough explanation of js interop? i'm having to translate stuff like app.$.foo.bar(baz) or Polymer.dom(document).querySelector("#foo .bar"), and it's mostly trial and error. the explanations i've found online tend to treat only the simplest cases.

(I mean, for somebody who is not a js ninja.)

:thinking_face: you could write a JS -> interop compiler lol

also is there some resource that has listed all the cljs react wrappers and what they are about

@rauh When I my CLJS compiles fine but I get a JS syntax error after I deploy my node app to the server because a library macro did something funny with a symbol, that's pretty obscure. I've already lost a lot of time on it.

I feel like never emitting illegal JS (but certainly sometimes emitting legal JS that doesn't run) is a perfectly reasonable line to draw for a JS compiler.

(or any compiler)

yah... i mean for now u could run an analyzer on the js after js build step

@mobileink is this insufficient? http://www.spacjer.com/blog/2014/09/12/clojurescript-javascript-interop/

that's my main resource, in fact.

for chained expressions like you share, the thing I use that isn't in that guide (as far as I saw) is -> which makes literal translations from examples easier

(-> app (.-$) (.foo) (.bar baz)) - the hard part is knowing when something is a property or method (.- vs. .)

and since js makes no such distinction, that requires either trial and error, console examination, or reading docs

but i find it assumes a little more js-fu than i have. e.g. sometimes js/foo.bar works, and sometimes it doesn't. looking for a little more hand-holding, i guess. 😉

ah, i was looking for sth like that!

js/foo.bar is sugar, and I don't really like it personally, it can always be replaced with (.bar js/foo)

ok, good, i don't mind trial and error as long as i know it's unavoidable.

separate issue: i've run into trouble with dashes and underscores. i understand aget and aset deal with this; what's best practice? always use aset/aget?

dashes should only come up for things defined for and by cljs itself, which shouldn't be accessed via aget / aset usually

wait, we should not write (defn foo-bar...) in cljs?

mind you this is all in the context of interop. my cljs must play nice with a js "object".

i'm delegating from a js object to a cljs ns.

When doing macros in a .clj file, how can I refer to, e.g. goog.object?

the js object is initialized via interop, clj->js.

@dominicm if you want to interact with goog.object at run-time, just use the symbol goog.object in your syntax-quoted form

@adamvh so use everything via full namespace symbol? Feels awkward, but okay 🙂

well, so what happens is that if your macro returns a syntax-quoted form, all the symbols in that form will be prefixed with the namespace in which the macro is expanded

after the macro is expanded

so if you have (ns my.cls.ns :require [goog.object :as go])

then (defmacro my-macro [] (syntax-quote (go/symb ...)))

should work how you want as long you expand it in a cljs namespace where goog.object is aliased to go

@adamvh That approach seems a little magical & prone to unexpected errors. I probably won't take that approach, but thanks anyway.

yeah, i think it's probably best to write macros that aren't couple to, like, expected namespaces in which they should be evaluated

An interesting reason this may not work in some situations (it does for me, but that's an aside), is if you're requiring, e.g. cljsjs.react in your macro namespace. You must add it to a cljs namespace of some kind. That's not terrible, but it does create an implicit coupling.

Can anyone help with a regex & run-all-tests? I'm trying to build up a regex and pass to cljs.test/run-all-tests. Here's the code:

(let [rgx (re-pattern (str "example" ".*"))]
  (cljs.test/run-all-tests rgx))
=> clojure.lang.ExceptionInfo: clojure.lang.Symbol cannot be cast to java.util.regex.Pattern

So it looks like the run-all-tests macro isn't using the runtime value, but the symbol? Any suggestions to pass a dynamic regex to run-all-tests?

Looks like, since run-all-tests is a macro, it is taking rgx as a symbol instead of taking its value, I suggest solving this by using a macro yourself

(I tested your case locally and found the same problem)

For example doing the let in the macro and then taking the value of the regexp inside the syntax-quote

cljs.user=> (let [rgx (re-pattern (str "example" ".*"))] #_=> `(cljs.test/run-all-tests ~rgx)) (cljs.test/run-all-tests #"example.*")