I’ve configured shadow to run with lein

I’m trying to release the code but I’m getting the error above.

@thheller OK, thanks. I'll try that.

It works all right now.

@wvelezva that error happens when using an invalid closure compiler version

make sure its [com.google.javascript/closure-compiler-unshaded "v20180910"] or later

since everyone else seems to be pasting their exceptions, I might as well participate too 🙂

sometimes during compilation we get this:

strange. shadow-cljs isn't even on the stack. not much I can do about that

yeah, I guess it has something to do with that hawk library.

yeah thats the file watcher on OSX

other platforms don't use that but the others are too slow on OSX

is that watcher activated dynamically based on the platform or should it be always there? I'm asking because I'm running linux 🙂

oh right. remembered wrong. only windows uses the default JDK watcher

but hawk also switches again and uses the JDK one for linux

okay, thanks for the info 👍 I think I'll report this to the hawk project, let them sort it out. in any case, the error doesn't seem to cause any issues, most likely it just fails to watch one or more files

hawk is more or less dead I think

no updates for years

problem is OSX since the JDK watcher is so slow. otherwise I wouldn't use it at all

but for linux I can probably filter it out and directly use the JDK one

yeah let me do that

wow, thanks 🙂

@lasse.maatta try 2.6.23. error might still happen but now I should be able to fix it if it does.

thanks @thheller, I didn’t know I had to add the closure compiler myself

anyone know if I can have a specific name for my release file?

right now I’m using a hook to the :flush stage to execute a “cp <outputfile> <releasefile>”

@wvelezva you know about the file name hashing support, right?

👀

no, sorry

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

if you’re renaming the file to avoid cache issues, you probably want that instead

otherwise I think it’s just the build name

I think you might be able to use a filename in :output-to, but I have not tried

that might just be for node

yes the :output-to it is only for node

rather..module name, is what I meant

yeah, module name…so using a diff kw as the module name will cause a diff name for the file

I got it, I just need two builds sections, one for my dev with the paths for dev

and a new one for my release

I don’t think that’s a great idea

why do you care about the filename?

when you run the release you get an optimized file

and code splitting is going to rely on those names, so…I think perhaps you’re asking for future trouble

I have not tried it, but you could perhaps set different output directories with https://shadow-cljs.github.io/docs/UsersGuide.html#_release_specific_vs_development_configuration

if you’re worried about “accidents” or something in deployment

it’s because we are used to use a name during development and then when the js file is complete we compilete it in the right production folder

I also test that and it “works”, but it is the same name for dev and prod

it is not really a problem

It’s just an old habit

😄

yeah, figured.

but I’m going to use it, it’s really simple that way

thanks @tony.kay

Hi there, Can someone refer me to an example of a shadow-cljs project with both cljs and clj files in order for me to understand how data can be required from one to the other? I basically want some to use some Clojure libraries in the back and serve the data in the cljs front.

@adamgefen not sure I understand the question. shadow-cljs generally focuses only on CLJS development and I recommend using lein for CLJ. as far as sharing sources is concerned you just point both at the same :source-paths and use the code as normal

@thheller so if I use some Java scraping library to grab xml to a clojure data structure, I should write it as edn to :source path and slurp it back in my cljs file?

no you probably want to create a HTTP server in CLJ for that and let the CLJS app talk to that

(assuming you are building something for the browser)

Ok. I'll look into that.

if you want something full stack I'd suggest looking at #fulcro

I want it for the browser but I did think that it will be useful to keep some logic in the back in order to be able to expand.

@adamgefen If you want the browser to convert xml into edn, you could go with an npm package such as https://www.npmjs.com/package/xml2js and then transform it to edn with js->clj.

thanks @grav, I'll have a look.

@thheller yeah, that was pretty old. Came in from here (what I started with to build the example): https://github.com/shadow-cljs/examples/blob/master/electron/package.json