Presented a talk on ClojureScript at the University of Cape Town yesterday during which we built a snake game. Slides, audio and gameplay now up: http://petrustheron.com/posts/sota-front-end-development-clojurescript.html Hopefully this is useful to others for teaching Clojure, especially the code (< 100 sloc).

This is nice, Petrus. Thank you for putting the slides, videos, and code together.

:thumbsup: thanks, @asyncsrc. Regretfully no video recording. Audio and slides unsynchronised, but I'll post compiled game soon

ah, I saw the screen recording videos within the slides of intellij, which was the most important part for me personally. It would be nice to get a video recording of the presentation, but in most cases it seems like i need to follow along manually with slides anyhow since they’re hardly ever in focus in the video. But yeah, this is definitely awesome. Thanks again

If i've got

(ns something.here (:require [goog.userAgent :as ua]))
(.log js/console (ua/getUserAgentString))

To answer my question: no

Also, what are some good resources for learning the parts of google closure that are useful from cljs?

Has anybody ever tried running ClojureScript on the server to return a pre-rendered single-page-application on the first request?

@pupeno: Reagent does this for it’s homepage IIRC

Oh!

what's a good option for a typeahead when you're already using Reagent and Bootstrap?

I'll use jquery-ui as described in the Reagent cookbook

maybe even better, reagent-forms

@borkdude: there’s some autocomplete stuff in goog + twitter’s typeahead.js has recently been added to cljsjs

hi @kolja!

@martinklepsch: thanks

In re-frame, how do I deal with a handler depending on another? Something like having a handler called get-user that request the user from the server, another got-user, that gets the data and puts it in the state, but now I have one called get-friends. Get-friends should only be run after get-user/got-user finished but I don’t want got-user to run get-friends as it’s not needed most of the time? Should get-friends do a dispatch-sync of get-user?

@escherize: for google closure: http://www.amazon.com/Closure-Definitive-Guide-Michael-Bolin/dp/1449381871

Thanks @greywolve - will check it out.

The latest Java update seems to have broken my cljs compiler. anyone else experience this?

with "openjdk version 1.8.0_51" everything works as expected

I’m using cljs.pprint but (pprint {:a {:b 1}}) does not generate any newlines.

I’ve bound print-pretty to true

tools.reader currently cannot read the number literal 0 in clojurescript. not sure if this is worth making a ticket since it’s not a stable release yet though.

@sekao: Under what circumstances? It’s tested here >> https://github.com/clojure/tools.reader/blob/master/src/test/cljs/cljs/tools/reader_test.cljs#L20 , and works for me.

What version of cljs/tools.reader are you using? What platform, etc.?

i’m using 0.10.0-alpha1, i’ll try again to see if i’m mistaken

i’m just doing (read-string “0”) and getting “Invalid number format [0]”. other digits work fine.

@andrewmcveigh: it must be an issue with my project because i can’t reproduce it in a new one

Yeah, not sure which read-string your calling, but (cljs.tools.reader/read-string "0”) is working for me.

Is it possible to include a figwheel live-reload server into an existing app?

@rickmoynihan: You want to compose figwheel into an existing process?

yes

I have an existing server project - with no client side js yet... I want to add client side clojurescript to that project

@rickmoynihan: oh yeah that is straightforward

I would take a peek at the quick start

and the readme

to understand the moving parts first

no template needed

so I've added lein-fighweel and lein-cljsbuild to the project.clj and put a cljsbuild section -- which appears to work (i.e. I can throw a cljs alert box up onto my page that is served when I run lein repl.

What I don't have figured out yet is the live reload workflow

Am I best doing something like this? https://github.com/bhauman/lein-figwheel#using-figwheel-from-the-repl

at first I would just use figwheel as a separate process

yeah that seems easier

ie. lein figwheel

when I tried it though it didn't seem to work

presumably I need to include a figwheel script onto the page?

you don't have to load your compiled assets from figwheel you can load them from the server.

when in the :dev env?

again I would look at the quick start

no real shortcuts

yeah I have compiled assets working

ok missed the quick start link - thought it was the README.md - thanks

shortcuts will drive you nuts

this looks like what I was expecting to do... thanks

yeah I should have given you the link, and not assumed that you were wanting a short cut

damn - turns I already had it working

lol

I hadn't seen the figwheel process had actually connected...

@yogthos: is it possible to generate in the reagent-forms typeahead in inside the list items an anchor? our bootstrap css expects that

@yogthos: maybe with result-fn?

@yogthos: Looks like that works

Nice :)

@borkdude: in general reagent-forms should be layout agnostic, it’s also possible to do multiple bind-fields calls binding to the same atom within a component, for example if you wanted to show different forms conditionally

@yogthos: great

@yogthos: the typeahead has a highlight-class

@yogthos: that works when you do arrow up and down, should it also work with mouse hover?

yeah that would make sense

would have to hook in mouse enters/leaves events

i think i found a way to lock up safari with cljs.js/eval by evaling a form defining a variadic function. if someone could try my test code in safari i would appreciate it https://github.com/oakes/boot-cljs-example/blob/master/src/app/core.cljs

@sekao: boot serve -d target/ watch speak reload cljs-repl cljs -sO none ?

i just use boot dev

ah cool. didn’t know there’s an alias in the example proj

@sekao: beachballs 😉

@martinklepsch: and if you remove the & and try it again it loads fine?

@sekao: yes — logs nil once

ok then i’m not going crazy 😃

I just created a re-frame project and it’s using figwheel but there’s no server side routing in the project. How do I make the server send the app when serving different URLs instead of just /

pupeno: easy way is to use # in your paths so everything gets to the same server path — otherwise, you’ll need to do some sort of path mapping — depends on server stuff how to do that, technically can just rewrite in apache etc.

nullptr: I just replaced secretary with silk and pushy to get rid of the #. In my main project, with compojure, it’s very easy to server the app no matter the URL and let the client do the routing. But for a blog post about this that I’m writing I’m using the re-frame template, which using figwheel with no server side, hence my question, how do I make figwheel serve the app for any URL that gets requested.

oh, in figwheel’s server you can pass it a ring handler which can basically do whatever it wants

Hey, I'm trying to set up modules. Right now I have

{:optimizations :simple
 :modules {
           :foo {:output-to "resources/public/js/foo.js"
                    :entries #{example.foo}}
           :bar {:output-to "resources/public/js/bar.js"
                       :entries #{example.bar}}}}

does example.foo :require example.bar?

nope

i figured it didn’t, though it would explain everything :)

@mattparker: I don't have any experience with modules, but if I remember correctly, in the example there are three parts: example.bar, example.foo, and common.code-to-both

and then there is a small dependency ordering

https://github.com/clojure/clojurescript/wiki/Compiler-Options#modules

@bensu: check this out http://swannodette.github.io/2015/02/23/hello-google-closure-modules/

yeah I remember that, but I'm just pointing you to the wiki in case the :depends option applies to your situation.

(that's all I can offer )

nah none of the namespaces depend on each other. Thanks though!

fwiw your expectations of the end result appear correct. at least that’s how closure modules have worked for me via JS build paths (plovr).

@sekao: There can be subtle problems with reading 0 related to *uncheked-if*. See https://github.com/mfikes/replete/issues/8#issuecomment-119052347

hi guys! I'm looking for a persistent collection for clojurescript that acts as a ring-buffer or sliding-buffer

any ideas?

so far I found https://github.com/amalloy/ring-buffer

but I would need to migrate that to cljc and replace a couple of things

@jpmonettas: Clojure (and cljs) has queues if that works for you

but I need a fixed length one

so if a push one more than supported size, it drops the first

something like core.async sliding-buffer

Gotcha, wasn't sure if you actually needed that.

exactly like amalloy ring-buffer but that works on cljs

@sekao: If you evaluate (do (ns foo.bar) (def x 3)) then x will not be defined in foo.bar in ClojureScript. In your “lock up Safari” example, that code complains a lot for me in things like Planck and Replete in the analyzer, perhaps as a consequence of the use of &. One suggestion would be to ensure cljs.user is created before doing the eval.

@jpmonettas: Closure has something (likely non-persistent). Never used it: https://google.github.io/closure-library/api/class_goog_structs_CircularBuffer.html

@mfikes i pushed a change to the example that evals them separately but Safari still locks up https://github.com/oakes/boot-cljs-example/blob/master/src/app/core.cljs

@mfikes the weird part about the (read-string “0”) problem is that it seems to work fine when i build with advanced optimizations, it only seems to fail when i’m in dev mode. but since i can’t reproduce it in other projects i am just ignoring it for now...

@sekao: When Safari locks up, does it use a lot of CPU?

yes, there is an item in activity monitor called Safari Web Content with high cpu usage

like 80% and it says it is not responding

There was a small perf problem we discovered with defn and Safari for cljs.js that can be worked around via :static-fns true: http://dev.clojure.org/jira/browse/CLJS-1381

@sekao: ^

@mfikes: yeah i noticed it is quite slow with defn in general but it seems like variadic defn is a separate issue, as it locks up indefinitely

@sekao: Yeah, I can repro your problem with :static-fns false in Planck.

@sekao (defn hello [& args]) ultimately kills it

oh maybe they are indeed the same issue, i think i got safari to lock up without the &

@sekao: It ultimately exhausted memory for me https://gist.github.com/mfikes/a6a35895cc38b9ab853c

@sekao: The good thing is the WebKit team already know about the defect (it has been known for a good chunck of a year.

i just cleared my cache and the non-variadic defn worked fine, so it may have been using my variadic test just now

that’s good to hear regarding the webkit team

@sekao: Planck (which is also based on JavaScriptCore) cannot evaluate (defn hello [& args]) unless I build it with :static-fns true to work around the JavaScriptCore bug.

i actually recall testing it on http://clojurescript.net and it worked fine in safari. i thought they were using bootstrapped cljs

Joel Martin updated http://clojurescript.net to reflecct what we learned about :static-fns true

oh ok

It was also exhibiting slowness for defn in Safari until he addressed that.

@mfikes: thanks, that indeed fixes the issue in my project

@sekao: Sweet. I updated https://github.com/clojure/clojurescript/wiki/Optional-Self-hosting

@jpmonettas: BOOM! https://github.com/amalloy/ring-buffer/pull/6. Hope it helps

@mfikes: I'll check that. Thanks!

@nberger: Hey, that was quick! Thanks a lot

@nberger: cool! I skimmed over the PR and I would recommend a project like ring-buffer to test on node (not only it's more likely to be installed but it's also a better fit)

@bensu: Thanks. Why do you think it's a better fit? Agree with the likelihood of node being available, but don't think it's something critical... Anyways, I'll wait until @amalloy reviews the PR.

@nberger: not critical at all. Node is faster (both upstart and runtime), more stable, and has fewer quirks than Phantom. The only reason not to use Node is when you need a dom to manipulate (which I'm guessing ring-buffer doesn't).

@bensu: well, that's not the only reason: another reason could be for example if you simply started using phantom because you consider it won't be much difference using node in that particular project. For example in this case, I'm really not worried by phantom being slower in running a test suite that takes 0.11 secs to run in my machine.

I agree, when I say the only reason I mean: "if you want to take the time to decide, node will be a better fit". In many cases, such as this one, the decision might not be worth it, as you just said

@bensu: I'd be very happy to change the cljs test config to node if the project maintainer considers it is important. Feel free to chime in the PR to add your point of view, that would be very helpful for the maintainer to take a decision.

@nberger: sure, I'll give it a go after the PR settles. the project looks cool and I've been thinking about something like this to implement a somewhat generic undo component.

thanks for porting it to cljs!

@bensu: cool! Go ahead, it has just been merged

@nberger: just did it, as you said the change is not worth at all.

@nberger: when you include the lein startup time, the difference is so small is in the rounding error

the upside is that it worked only by changing the compiler's config, which is always a relief for a library author

@bensu:

anyone happen to be really familiar with running cljs tests via figwheel?

I’ve got some really weird behavior

I’m sure it’s a flaw in my configuration

but I don’t have my head around how it all works well enough yet to get it to go through

the short of it being that whenever I edit a non-test file to add new symbols they get added but an old version of the tests is run against them

and whenever I update the tests the newest test version is run but against old symbols

I’m running it like lein figwheel app test

atop something not dissimilar from the reagent template configuration

and I think what must be going on is that the two builds are conflicting somehow

but I can’t really figure out how or why

@tel: faced a very similar problem with boot-cljs; i created a simple repro here: https://github.com/pandeiro/cljs-test-issue

is cljs.test preferred over cemerick.clojurescript.test?