ok this is really nice to work with

😄 we really like it. been using it at work for about 5 months since the start of our project

The shadow-CLJS manual is 20000 words long

so much left to add still 😞

@thheller Draw a picture. A picture says more than 1000 words. Or even better: upload an interpretive dance! I’m sure that says more than 20000 words.

I wish 😛

I’m having trouble with release building of a node library target. Getting

TypeError: a.Ga is not a function
extensionHostProcess.js:452
    at Object.ih [as greetings_activationGreetings] (/Users/pez/.vscode/extensions/cospaia.clojure4vscode-1.3.33-TS-CLJS/out/lib/calva.js:276:520)
    at activate (/Users/pez/.vscode/extensions/cospaia.clojure4vscode-1.3.33-TS-CLJS/calva/extension.ts:114:14)

https://code.thheller.com/blog/shadow-cljs/2017/10/15/externs-the-bane-of-every-release-build.html

@pez do you have any other JS code on the page? nvm you’re using node

https://shadow-cljs.github.io/docs/UsersGuide.html#externs

@pez if you prefer not to deal with externs you can set :compiler-options {:optimizations :simple} although I do recommend trying :compiler-options {:infer-externs :auto} first

Thanks! (I was looking for where functions in my library were not properly exported… 😃 ) I could get past the first hurdle by adding a type hint. Now will see if :infer-externs does the trick. I should get less of these problems once I have factored the code better between what I do in TypeScript code and what I do in ClojureScript code.

@thheller: I will be moving Calva development to implementing more and more of the “business rules”/logic in CLJS and do the integration in VSCode in TypeScript. (Building the whole extension using shadow-cljs worked, but meant that I couldn’t use the debugger for all that JS code that still remains to be ported and I can’t afford the slowdown in development that that brought.) Anyway, so I will build the logic as a node library using shadow-cljs. Thing is the library API will get pretty “fat”/have many functions. Since namespacing is dropped, I am now trying with prefixing the function names. Is there a cleaner way, that you can see?

you can use :npm-module so the TS part can import namespace separately when required

Wow, seems the convenience mode should work for me.

well not if you plan on deploying the code anywhere

well, if you publish under your own package name that can do it of course

The code will be deployed anywhere the Calva extension is installed, so I will have to learn some about building npm modules, I guess. But the name spacing it allows is so much nicer than what I have now, so I’m happy to investigate.

@pez you can just make it part of your calva stuff

its basically just a directory you need. it doesn't have to be entirely separate package

I'm trying out Cursive. And I decided to also try out :node-script at the same time, since I just want to play around with ClojureScript and Cursive's REPL integration. I've written a minimal shadow-cljs.edn, ran npx shadow-cljs pom, left a running npx shadow-cljs server in a terminal, imported pom.xml into Cursive, connected to the REPL, I am able to do (shadow/watch :the-id) and and then (shadow/nrepl-select :the-id) - Cursive recognizes it at that point as a CLJS REPL. But if I try to evaluate anything - I get "No application has connected to the REPL server. Make sure your JS environment has loaded your compiled ClojureScript"

So I guess :node-script isn't the best option for that kind of experimentation, do I really need to switch to :browser and keep a browser running even though I just want to play around with the language itself?

just use (shadow/node-repl) instead

otherwise you need to actually run the node script manually. ie. call node the-script.js

oh I see. I read that section of the manual, but had a brain fart and didn't think it applies to me. In that case I don't even need to tell shadow to watch the app, since I'm sending code to the repl via cursive and not running the compiled output, right?

node-repl doesn't watch yes. :node-script does.

REPL command timed out

let me restart Cursive just in case

what command?

just what Cursive output when I told it to switch the REPL namespace to the current file's one

and shadow-cljs shows [:did-not-find myns.core]

Just to be sure I'm doing it right: generate pom, import pom into cursive; start shadow-cljs server; start cursive nrepl connection; issue (shadow/node-repl) - at that time cursive shows that it is in CLJS repl. but then setting the REPL namespace fails

did you require it first?

yeah I can reproduce with (in-ns 'ns.that.wasnt.loaded-yet)

guess cursive just sends a blank in-ns

I tried loading the file in the repl (via Cursive's REPL / Load file in REPL) and got the following:

Loading src/myns/core.cljs... repl/require failed { Error: ENOENT: no such file or directory, open '/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/clojure.set.js'
    at Object.openSync (fs.js:443:3)
    at Object.readFileSync (fs.js:348:35)
    at global.SHADOW_IMPORT ([stdin]:45:15)
    at Object.shadow$cljs$devtools$client$node$closure_import [as closure_import] (/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/shadow/cljs/devtools/client/node.cljs:33:4)
    at Object.shadow$cljs$devtools$client$node$repl_require [as repl_require] (/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/shadow/cljs/devtools/client/node.cljs:61:7)
    at shadow$cljs$devtools$client$node$process_message (/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/shadow/cljs/devtools/client/node.cljs:122:6)
    at Object.shadow$cljs$devtools$client$env$process_ws_msg [as process_ws_msg] (/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/shadow/cljs/devtools/client/env.cljs:148:10)
    at WebSocket.<anonymous> (/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/shadow/cljs/devtools/client/node.cljs:159:12)
    at WebSocket.emit (events.js:182:13)
    at Receiver._receiver.onmessage (/Users/bupkis/dev/myns/node_modules/ws/lib/WebSocket.js:141:47)
  errno: -2,
  syscall: 'open',
  code: 'ENOENT',
  path:
   '/Users/bupkis/dev/myns/.shadow-cljs/builds/node-repl/dev/out/cljs-runtime/clojure.set.js' }
done

if I require it manually in the REPL - loading the file afterwards works fine

not a clue whats happening. the recent nrepl change seems to have messed up something on my side and I can't use nrepl anymore

god dammit I hate nrepl so much

I wonder if connecting from Cursive to the socket-repl with something like tubular would be better

not sure if Cursive supports socket repl at all

you can create a clojure.main config and set parameters to -m shadow.cursive-repl

but tubular goes around that

that will automatically connect to the running shadow-cljs server

let me try that

[2:1]~cljs.user=> (require 'demo.script)
nil
[2:1]~cljs.user=> (in-ns 'demo.script)
nil
[2:1]~demo.script=> 

this works for me. thats all I know 😛

yeah if I require it first, then load-file works

btw. the clojure.main with -m shadow.cursive-repl raises an exception

https://pastebin.com/xbXkAHQy is the stack trace

thats all?

the only interesting bit is missing 😛

yeah just noticed cursive didn't copy it

had to expand manually

at java.util.Formatter.checkText(Formatter.java:2579)
	at java.util.Formatter.parse(Formatter.java:2565)
	at java.util.Formatter.format(Formatter.java:2501)
	at java.util.Formatter.format(Formatter.java:2455)
	at java.lang.String.format(String.java:2940)
	at clojure.core$format.invokeStatic(core.clj:5684)
	at clojure.core$format.doInvoke(core.clj:5678)
	at clojure.lang.RestFn.invoke(RestFn.java:423)
	at shadow.cursive_repl$get_socket_port.invokeStatic(cursive_repl.clj:25)
	at shadow.cursive_repl$get_socket_port.invoke(cursive_repl.clj:10)
	at shadow.cursive_repl$repl.invokeStatic(cursive_repl.clj:31)
	at shadow.cursive_repl$repl.invoke(cursive_repl.clj:29)
	at shadow.cursive_repl$_main.invokeStatic(cursive_repl.clj:93)
	at shadow.cursive_repl$_main.invoke(cursive_repl.clj:87)
	at clojure.lang.AFn.applyToHelper(AFn.java:154)
	at clojure.lang.RestFn.applyTo(RestFn.java:132)
	at clojure.lang.Var.applyTo(Var.java:702)
	at clojure.core$apply.invokeStatic(core.clj:657)
	at clojure.main$main_opt.invokeStatic(main.clj:317)
	at clojure.main$main_opt.invoke(main.clj:313)
	at clojure.main$main.invokeStatic(main.clj:424)
	at clojure.main$main.doInvoke(main.clj:387)
	at clojure.lang.RestFn.applyTo(RestFn.java:137)
	at clojure.lang.Var.applyTo(Var.java:702)
	at clojure.main.main(main.java:37)

hehe oops. never had that error so I never noticed

are you in the correct directory? and is the shadow-cljs server instance running?

the error above is just a missing .shadow-cljs/socket-repl.port file which should exist when the server is running

yes the file is there and the port is the correct one. the run configuration for that clojure repl has the working dir set correctly

oh hehe . that script is really old

you can set :cache-root ".shadow-cljs" in your shadow-cljs.edn

I'll modernize the script a bit

yeah that works

but I think I'll stay with the nrepl way and manually requiring the NS before trying to load the modified file in the repl

the autocomplete is busted in the cursive-repl mode

but that's probably cursive's fault

I'll try to figure out what happened to my nrepl stuff

then I can test whats actually going on

I know you're busy so no worries, the node-repl thing is a working solution

btw. is :node-script the best way to do what I'm trying to - e.g. to have a live repl in Cursive that I can load files into and play. It's not a problem to run some Node process manually outside of Cursive if need be, I'm already running the server so one more won't hurt hehe

no node-repl is better for that

unless you also want the node script to actually do something itself

if you just want the REPL then node-repl

thanks 🙂

thheller, how hard do you think it would be for someone to carve out a pared down version of shadow-cljs that just built code? as in no watcher, repl, etc.

I think it would be very useful for running on e.g. a toaster jenkins or gitlab runner

whats wrong with shadow-cljs compile?

that doesn't start any watchers or anything?

IME when running on CI, starting the shadow-cljs process to run shadow-cljs release app takes the majority of the time spent on that task

thinking of ways to optimize the startup time

you never reported back with any actual startup times

did you address the caching issues?

I did address the caching issue, and that sped up the building of the code itself a bit. but the startup can still be in the ~3-5 min range

do you have shell access to the box running the stuff?

obviously it gets worse the more jobs that are running. we’re working on getting beefier CI servers and some other enhancements

probably not; it’s a docker container running I’m-not-sure-where tbh

are you using deps.edn yet?

not at work, no. we’re using leiningen, but I didn’t notice any changes between using regular shadow-cljs and the leiningen integration

before you try to optimize anything in the dark you should really gather some data first

agreed. I’m on my day off and thinking of hare brained ideas 😛

launch a blank lein repl and then (time (require 'shadow.cljs.devtools.api))

if that actually takes 3-5min we can talk about tuning that

I can tell you that running the container locally on my 16gb macbook, the build times are completely reasonable

5min seems pretty excessive to load some code

but not joking I have time npx shadow-cljs release <builds> running on our CI and I easily see ranges from 5-10mins

I totally believe you but I would still like to see a npx shadow-cljs release <builds> --verbose report

I need to know WHERE the time is spent before thinking about optimizing anything

there is a lot of stuff happening even on startup

so it could be the classpath indexing just being terribly slow

totally. I’ll gather some data and report back early next week

my main task for this sprint is improving our CI 🙃 so if there’s some low hanging fruit I can grab once we know more I’ll be happy to do the work

so if you cache only .shadow-cljs/builds you might want to add .shadow-cljs/jar-manifest

but again ... without any data this is impossible to make suggestions for

assuming you actually fixed the cache issue

we are persisting .shadow-cljs/builds and node_modules at the moment - not sure about jar-manifest. I can’t remember if I only kept the builds folder or the whole .shadow-cljs

I’ll check on monday 👍

its only a guess. also the disk performance really matter. just the other day @caleb.macdonaldblack changed something in his container setup to go from 3sec+ recompile times down to 0.3secs

so thats another thing to look out for

but really run with --verbose. the output prints just about everything where any time is spent

well its probably still missing a lot of details but it should help

@thheller new here, just wanted to say thanks for all the work you've done on shadow-cljs! I've been waiting so long for a viable dev setup for working with ClojureScript in a "normal" JavaScript project

welcome. glad you like it.

it's a relief to not have to deal with leiningen for a project that doesn't use any Clojure code on the JVM

Has anyone been able to set up rebel-readline in the shadow-cljs REPL? Not sure if it's incompatible with how shadow-cljs works, but it'd be awesome to have that readline experience

I have not looked at it yet. the CLJ repl should work out of the box. no idea how to hook up the CLJS REPL though

Yeah, that's the part I wasn't sure about

I always recommend hooking up your editor first. I only use the REPL from Cursive the readline was never important to me

I've gotten shadow-cljs to work with Calva in VS Code, but the REPL experience isn't that great there. I'll try setting it up with ProtoREPL next

https://medium.com/@jacekschae/slick-clojure-editor-setup-with-atom-a3c1b528b722

Yep, found that one last night, thanks!

I suppose I should try to get it working in Atom since I'm a member of the Atom team 😉

feel free to open an issue about rebel-readline

Yeah, I think I'll try to give it a shot and set up a demo repo if I can get something working

the question has come up multiple times now so if there is an issue I maybe won't forget about looking at it at some point 😉

@daviwil: as the author of Calva, I’d like to know what about the REPL experience isn’t great in your opinion. I have tons of ideas of where to focus my efforts, but time is limited and knowing what would make people stick with Calva is gold. #calva-dev is a good place to talk about these things. 😃

about to add select-nrepl support in cider 😉 feel free to try it out https://github.com/clojure-emacs/cider/pull/2397

about the above, I wonder whether a watch-all could be a good idea ...

It looks like http://shadow-cljs.org is down?

Timing out for me too

^ @jiyinyiyong

the domain is maintained by @jiyinyiyong and hosted in China