How do I access session data from the clientside?

ring-session

Is it possible to access ring response map parameters from Clojurescript?

My solution is to pass ring-level params on the :body key of the response map. Oh yeah. Include it in the body. das rite. can't believe i didn't land on this solution earlier. i feel like i knew it but just kept assuming there was a different way. javascript can access dom elements and therefore including data in the body actually makes sense.

I just verified this solution. I am extremely happy with the results. My clojurescript application can have immediate access to session variables via :body embedding and I can have SEO without the need to serverside render much.

what's the easiest way to inject a CLJS REPL into an existing webpage? I'm trying to debug some JS interop stuff and it would be way quicker to explore with the REPL ...

Hello! I have a generic clojure question in regards to building a clojurescipt project - take a look at this function:

(defmethod start-cljs-repl :nrepl
  [_ figwheel-env]
  (try
    (require 'cemerick.piggieback)
    (let [cljs-repl (resolve 'cemerick.piggieback/cljs-repl)
          opts' (:repl-opts figwheel-env)]
      (try
        ;; Piggieback version 0.2+
        (apply cljs-repl figwheel-env (apply concat opts'))
        (catch Exception e
          ;; Piggieback version 0.1.5
          (apply cljs-repl
                 (apply concat
                        (assoc opts'
                               :repl-env figwheel-env))))))
    (catch Exception e
      (let [message "Failed to launch Figwheel CLJS REPL: nREPL connection found but unable to load piggieback.\nPlease install 
        (println message)
        (throw (Exception. message))))))

@voltecrus I'd say yes. catch Throwable e is a bit more generic but in practice it rarely matters

is there anything not working like you'd expect?

@vemv that's the thing. I'm trying to setup chromex with figwheel/cider but it seems that i cant get the figwheel properly setup through nrepl because whatever error gets caught there and I became stuck. Decided to look into inferior-clj because everything seem to work if you drop the nrepl/cider combo

if anyone’s using emacs, how do you connect re-frame app to CIDER? I’m getting loads of errors in cljs 😞 in a freshly generated re-frame template

it can’t find namespaces whenever I evaluate some piece of code

@bravilogy Too vague to troubleshoot, see response in #cider

i'm using secretary for routing and it's not changing the browser url whenever i use the secretary.core/dispatch! function to change the route

is this how is it supposed to work? in case it is, do you guys have any recommendation on how to change the url whenever i change the route? is it something simply not supported by secretary?

i thought it would be using window.history.pushState internally for that, but it doesn't seem to be

I think you're looking for https://github.com/venantius/accountant

thanks @joelsanchez, it seems to be solution!

glad I could help, I'm using it with bidi and it works flawlessly so far

(cljs-noob question) I'm trying to get SQLite working with an Electron app, but I cannot seem to figure out how to use it properly. I'm not seeing any successful examples online, so I thought I'd ping here to see if folks can help. I'm using loom as my REPL to test things. Here is a gist of what I'm doing right now: https://gist.github.com/defndaines/58277fb975475ac2f522a2b1f1b849c8 I cannot seem to get past the SQLITE_RANGE: bind or column index out of range error. I'm comfortable with Clojure, but new to CLJS, node, electron, et al.

(clj->js [id]), probably

most likely it's interpreting your [id] as a cljs collection, which is an object in JS land, hence trying to treat it as named parameters (as per the documentation https://github.com/mapbox/node-sqlite3/wiki/API#databaserunsql-param--callback )

I think #js[id] would likely be better, but yes

Thanks, @joelsanchez That does the trick. Is there a reason to favor clj->js over #js, other than bevity?

clj->js is fairly complex and tries to do a recursive transform

I favor clj->js because #js only transforms one level

see - I prefer #js because I always want the weaker tool

is there a function named #js? i couldn't invoke it

it's a reader macro

@joelsanchez reference here - sadly the > gets escaped, search for clj->js https://clojurians-log.clojureverse.org/clojurescript/2015-12-08.html

tl;dr thepeople that wrote clj->js are skeptical of it and think you should avoid using it whenever possible

dnolen 16:08:12
clj->js is just convenience - it performs terribly and has many corner cases

interesting! thanks

is it possible take a cljs form and, from the repl, emit the unoptimized js it will compile to?

@hagmonk There are at least 3 ways to do this. • You can enable :repl-verbose See https://clojurescript.org/reference/repl-options#repl-verbose More on that option here http://blog.fikesfarm.com/posts/2015-06-15-see-js-in-cljs-repl.html For quick tests, lumo -v or planck -v is probably the quickest way to achieve this. • If you don't want to restart your REPL, (set! *print-fn-bodies* true) and then evaluate the form you want in a function literal like (fn [] (+ 2 3)) • Use the self-hosted compiler at the REPL and compile the string form: (require 'cljs.js) and then (cljs.js/compile-str (cljs.js/empty-state) "(+ 1 2)" nil {:eval cljs.js/js-eval :context :expr} identity)

The 2nd way is the way you want to go if your form depends on your REPL environment.

@defndaines In addition to the inherent function call, (clj->js [1]) compiles to fairly verbose (unoptimized) JavaScript:

cljs.core.clj__GT_js.call(null,new cljs.core.PersistentVector(null, 1, 5, cljs.core.PersistentVector.EMPTY_NODE, [(1)], null))

[(1)]

If, at runtime, you have a persistent data structure you'd like to convert to JavaScript, clj->js is cool, but if you essentially have a compile-time literal, #js is arguably better.

If performance is critical at the point you are doing the interop, here is a relevant post, covering both directions: http://blog.fikesfarm.com/posts/2017-11-09-avoid-converting-javascript-objects.html

Bit struggling with equality checks on deftypes

(deftype AudioSignal [val])
(let [a AudioSignal
      b (AudioSignal. 1)]
  (case a
    AudioSignal (prn "yo")
    (prn "no"))
  (case (type b)
    AudioSignal (prn "yo")
    (prn "no")))

On the other hand, this returns true (= AudioSignal AudioSignal) => true

@hlolli Consider the behavior of this: (let [a 'AudioSignal] (case a AudioSignal :y :n))

hmm ok, this succeeds

The case test constant for that example is a symbol.

I wonder how that quote tesest against when the compiler adds the namespace in front of the typename.

(See the last paragraph of the docstring for case.)

`The test-constants are not evaluated. They must be compile-time literals, and need not be quoted.` lol ok

In fact @hlolli (let [a 'AudioSignal] (case a AudioSignal :y :n)) has nothing to do with your deftype. It is just testing symbol equality

yes I see, the errors I was getting at one point were

No matching clause: function (val){
this.val = val;
}

so ok, I didn't know this about case, never tripped on this before

@hlolli There is a quirk in ClojureScript, related to ^:consst, to be aware of, which was probably a mistake and doesn't work in Clojure. More about that here: http://blog.fikesfarm.com/posts/2015-06-15-clojurescript-case-constants.html

nice thanks @mfikes!

@hlolli Knowing this, perhaps this is what you want:

(deftype AudioSignal [val])
(let [a AudioSignal
      b (AudioSignal. 1)]
  (condp = a
    AudioSignal (prn "yo")
    (prn "no"))
  (condp = (type b)
    AudioSignal (prn "yo")
    (prn "no")))

(In short condp = for "runtime", case for "compile-time" might be the best way to look at that.)

hmm, yes, I guess I'll go trought the booleans again a bit. But case seems more "extreme" macro than cond family.

knowing it becomes js switch then it makes sense.

another possibility, other than condp =, is to simply use a map

case is awesome for two reasons, IMHO: 1) It is very easy to read when the tests are compile-time constants. 2) You get constant-time dispatch. In other words O (1) instead of O (n)

yup, in my case I'd go for symbol quoting, have many types I made for a transpiler. But I'm on new terretory, experimenting with stuff I didn't test before..

Wow, assuming these benchmarks are valid, if you want to do a mapping in code where the mapping is known at compile-time, it looks like case is a lot faster than using a map literal, with these speedups V8: 30, SpiderMonkey: 18, JavaScriptCore: 42. (This means 18 to 42 times faster, under :advanced)

(defn lookup-using-map-literal [k]
  ({:a 1 :b 2} k 3))

(defn lookup-using-case [k]
  (case k :a 1 :b 2 3))

(simple-benchmark [f lookup-using-map-literal] (f :a) 1e8)
(simple-benchmark [f lookup-using-case] (f :a) 1e8)

@mfikes thanks for that! just catching up on your comments now