Hi! I have an issue where my cljs build now produce a goog not defined error, it started after I did a lein clean

the answers I found on google apply to shadow.cljs but I don't use it, and not sure how to fix it

I made sure I use the latest versions of clojurescript and cljsbuild

^ fixed it by re-enabling whitespace optimization in project.clj but that feelswrong

@archibald.pontier_clo do you use a build config without a :main or :modules entry?

it has a :main entry

are you loading the correct file? do you get any 404s when loading the page?

@thheller sorry for the late reply, no 404, and yes I am loading the correct file, enabling whitespace optimization in project.clj stopped the error without any other change in the code

Is there any reason not to use syntax-quote at runtime? The docs say “Only intended for use in Clojure macros” (http://cljs.github.io/api/syntax/syntax-quote) but I’m not sure why. eg: https://www.maria.cloud/gist/c2282b77e2e47d71cb68b3becce8f84c?eval=true

I suspect it may have issues with properly namespace qualifying symbols in certain nested scenarios, as namespaces or vars don't exist at runtime in (non self-hosted) cljs

so I wouldn't say it's "only intended for use in macros", but with that gotcha

those are some very good docs btw, good job @shaunlebron

that is true. that is also a problem when using syntax-quote in macros, as the clj namespace doesn’t know about vars in cljs namespaces.

@mhuebert it’s mainly a guidepost for me to say “syntax-quotes are built for macros and nobody uses them otherwise and perhaps they shouldn’t be”

@shaunlebron that's not true tho

using there's nothing tying syntax-quote to macros

and I do use it at runtime a fair bit :P

sweet, how?

the entire tools.emitter symbolic bytecode is expressed through syntax-quoted templates

that's in clojure, not cljs, but doesn't really make a difference

`[:li ~@(when condition? [:li "foo"] [:li "bar"]) [:li "baz"]]

https://github.com/clojure/tools.emitter.jvm/blob/master/src/main/clojure/clojure/tools/emitter/jvm/emit.clj#L214-L230 as an example

someone posted an example like that earlier (I think in #clojure ?)

that was me :)

I think that works just fine in CLJS too

so you’re creating a defmethod where the body’s code is generated at compile time, or rather evaluate time?

this is making my brain fold

I might ping @mfikes to see how much of this is possible in cljs before amending the docs

no no syntax quote has nothing to do with compile time or runtime

it's complicated because in Clojure you don't think about it as "compile time"

this has nothing to do with evaluation differences in clojure or cljs

it's the exact same

typing

`[~@(when x whatever) 1 2 ]

(vec (concat (when x whatever) (list 1) (list 2)))

at runtime there absolutely no difference

wow, just like matt’s example in maria, i did not know this

as a protip

take a syntax-quoted expression

stick a quote over it

and you'll see what it translates to at runtime

user=> '`[~@(when x whatever) 1 2 ]
(clojure.core/apply clojure.core/vector (clojure.core/seq (clojure.core/concat (when x whatever) (clojure.core/list 1) (clojure.core/list 2))))

ok, it's not literally identical to what I pasted, but it's equivalent :)

i need to stare at this for a minute

too many questions

in different directions

how do i trace what’s going on

i inserted references to the reader code in the docs, maybe i have to look at them more closely

think about this in clojure not clojurescript, trust me that in this case there's no difference and it'll make it a bit easier to grasp

then sidestep the repl and play the repl manually using eval and read-string

user=> (read-string "`[~@(when x whatever) 1 2 ]")
(clojure.core/apply clojure.core/vector (clojure.core/seq (clojure.core/concat (when x whatever) (clojure.core/list 1) (clojure.core/list 2))))

user=> (read-string "'`[~@(when x whatever) 1 2 ]")
(quote (clojure.core/apply clojure.core/vector (clojure.core/seq (clojure.core/concat (when x whatever) (clojure.core/list 1) (clojure.core/list 2)))))

this is helping: https://github.com/clojure/tools.reader/blob/tools.reader-1.3.0/src/main/clojure/clojure/tools/reader.clj#L693

this makes it clear why the quote trick works

that helps too, yeah

yes, the thing to grasp I guess is that syntax-quote delegates the templating at runtime

and it couldn't do it any other way!

otherwise you could only syntax-quote forms containing constant values, which wouldn't be very interesting

when somebody says "x only works in macros", they forget that a macro is just (big handwaves here) (eval (the-macro-as-a-function args))

so if it "works in a macro", it must also "work" in a function, at runtime

because macroexpansion time is the runtime of a macro

and a macro is just a function

hope that makes sense

yeah that helps, my understanding is better

hard to keep the evaluation differences between clj and cljs in my head and to see how that affects this

it doesn't, modulo name resolution in syntax-quote

so it does, since that can certainly affect it

but even that is not a problem unless you're using syntax-quote inside a read-string

well, and that's a big if. if name resolution doesn't work as expected, then at the least I would want to understand exactly how

like resolving vars at runtime i think

yep but see my last caveat

ah

so in clojure you can do

(read-string "`+")

because the namespace map is available at read-time being executed at runtime

that is different than doing simply

`+

read-time at compile time in clojurescript has no differences than in clojure

it can get really confusing when you can nest stages like this (`read-string` is nothing but read-time available a-la-carte at runtime!), but this is what makes lisps so powerful :)

i think i need to hold on to the issue of name resolution, i don’t understand how read-string affects that

we’re talking about cljs now

cljs has read-string

either cljs-js or cljs-jvm

I'm talking about cljs-jvm

yeah, read-string is the runtime read-time thing

self hosted cljs does have reified namespaces and var I think so that makes it similar to clojure (but I never used it so I'm not sure)

but is read-string being invoked when syntax quotes are read you mean?

what is "it"?

let me backtrack a bit

you can think of (non-self hosted) clojurescript as: (cljs.emit/emit (cljs.analyzer/analyze (clojure.tools.reader/read-string <..>))) yeah?

this is all happening on the jvm, at compile time

let's say <..> is

"`+"

tools.reader on the jvm reads that, interops with the namespace map of the clojurescript analyzer which is available at compile time, and returns cljs.core/+ to the analyzer

but if you have

"(read-string \"`+\")"

that read string is pure runtime

and is cljs.tools.reader/read-string

which does not have access to the namespace map

and so can't perform name resolution

wow yeah, this makes sense so far

ok, now all the examples that I shown you before can be explained and understood by replacing <..> with them :)

it's really tricky to keep it all in your head but when it clicks it's really simple

graphically it's much easier to follow

> but even that is not a problem unless you’re using syntax-quote inside a read-string

ok, i’ll try to come up with an example to put in the docs for this

:thumbsup:

thanks for that tour, crazy and helpful!

haha you're welcome

https://github.com/cljs/api/issues/180

(fn [cb]
      (let [id (atom -1)
            frame (fn [] (reset! id (requestAnimationFrfame
                                     (fn []
                                       (cb %)
                                       (frame)))))]
        (frame))

Having trouble running this code. I get an error that (frame) call is undeclared

Hey I'm trying to write a unit test with this clause: (is (thrown? js/Object (ig/unique #{:just-me :no-theres-me-too!})))

But it's failing:

FAIL in (utility-test) (cljs/test.js:373:11) (thrown? js/Object (ig/unique #{:no-theres-me-too! :just-me})) failed with

I think you need a ( in front of function call

Otherwise, it is a parameter to a function

(is (thrown? js/Object (ig/unique #{:just-me :no-theres-me-too!})))

@jayzawrotny as far as I am aware let in cljs is a let*. This means that the name frame is not bound to a value until the expression (which references frame) is evaluated.

Do you see the circularity in that?

@eric.d.scott what's ig in this case?

The namespace where the function 'unique' lives. Expects a singleton, and throws an error otherwise

have you tried using js/Error instead of js/Object?

Print out the entire e in the console

I am sure the exact namespace is inside var e

str is just giving you a browser agent header

Yes I actually started with js/Error, then saw cases where people claimed js/Object was the way to go.

Javascript doesn't use a class inheritance system... its a prototypal

Not everything inherits from Object in JS

Like Ruby does... where everything is an Object

I will verify that, right now, for you

here is an example, to find out who is the parent of an object, e being the object in this case…

set hidden field message_receiver_id = 2 var t = “” undefined t.isPrototypeOf ƒ isPrototypeOf() { [native code] } t.isPrototypeOf ƒ isPrototypeOf() { [native code] } t.instaneof undefined t.instanceof undefined typeof t “string”

We don’t recommend catching errors on Exception in Ruby because it will catch all errors, when the test is targeting very specific errors

@eric.d.scott I don't understand why js/Object is preferred here, and would probably disagree w/

claims otherwise

for catching anything there's (catch :default e ...)

@dnolen loved the talk on clojure.spec

👍 👍 👍

hi guys, sorry if stupid question but i worked sometime now on clojure but i can’t say i am exactly a guru 🙂. So i have this regular expression in a cljc file #"(?i)<\/?((script|style|input){1})(\s?[^<>]*)>" in the js file it becomes: /<\\/?((script|style|input){1})(\s?[^<>]*)>/i that… 🎉 gives me this error:

var x = /<\\/?((script|style|input){1})(\s?[^<>]*)>/i
VM1576:1 Uncaught SyntaxError: Unexpected token {

am i doing something wrong?

in clj it works very well, this file has been a clj file for some time now and it always worked, i turned it into a cljc and i got this problem in js

if you wonder:

[org.clojure/clojure "1.10.0-alpha7"]
[org.clojure/clojurescript "1.10.238"]

https://stackoverflow.com/questions/36775526/regex-patterns-in-java-vs-javascript seems relevant

thanks, i did some googling but i couldn’t find anything, probably wrong keywords

Issue: cookies don’t save http://dev.localhost:8080/authentication - ring server http://dev.localhost:9500/log-in - re-frame SAP I send JSON from SAP to server by https://github.com/Day8/re-frame-http-fx, cookies don’t save in web browser If i run directly http://dev.localhost:8080/authentication cookies save in web browser I added

(-> (handler request)
        (header "Access-Control-Allow-Origin" "")
        (header "Access-Control-Allow-Credentials" "true")
        (header "Access-Control-Allow-Headers" "Content-Type, User-Agent"))

(wrap-session {:store (jdbc-store postgres/db {:table :sessions})
                     :cookie-name "auth-session"
                     :cookie-attrs {
                                    :http-only false
                                    :same-site :lax
                                    :secure false ; https
                                    ;:domain "dev.localhost"
                                    ;:max-age 1800
                                    }})

When I add [:img {:src ""}] cookies save so it is about https://github.com/Day8/re-frame-http-fx and goog.net.XhrIo (`XMLHttpRequests`) queries

yeah… I missed :with-credentials true in ajax request…

*I don’t recommend to set all this options in server for security reason. I set them to find the issue.

@nathantech2005, @dnolen Thanks for your help. I finally wound up going with the following:

(defn erroneous-unique [] "Couldn't get (thrown?...) to work" (try (ig/unique #{:just-me :no-theres-me-too!}) (catch :default e :error))) (deftest ... (is (= erroneous-unique) :error) ...

@eric.d.scott I am not in topic, but why not (is (thrown? …)) ?

There are also fn to read data from exceptions

https://clojuredocs.org/clojure.core/ex-info

am I allowed to ask a question about raw Javascript? We can thread it. I am looking for a send method in JS… to dynamically make a method call on an object.

var method = "log";
console[method]("bob");
bob

@kwladyka - the problem was getting the throw? clause to key on the error class. Nothing I tried made it happy, and the test kept failing.