FROM circleci/clojure:openjdk-11-tools-deps-1.10.0.414-node

My :modules config has this form

{:base {:entries [cljs.core]}

 :mod {:entries    [my.ns]
       :depends-on #{:base}}

 :worker {:entries    [my.ns.worker]
          :depends-on #{:base}
          :web-worker true}}

@UKQS17P3L set :devtools {:browser-inject :mod}

I'm also in trouble with react-native target trying to access js/document. But I downgraded fulcro with shadow-cljs at once because I run out of time (so I can't be sure about shadow-cljs or fulcro issue).

@thheller it worked, thank you! Is this documented somewhere? I am usually referencing this guide https://shadow-cljs.github.io/docs/UsersGuide.html

no I still need to document it. I added it because previously workers didn't support hot-reload but now they do

Yes I saw that 🙂 pretty nice

thought nobody was using web workers so I didn't bother documenting it yet 😛

No problem, shadow keeps improving and it is what matters

I haven’t had any success with this so far, if anyone finds anything - please let me know 🙂

this is the char 🠶?

That appears as a box as well.

https://www.htmlsymbols.xyz/unicode/U+1F836

A page of boxes 🙂 I guess Ubuntu 19.10 does not have that built in?

guess not 😛

although there is an ubuntu at the bottom 😛

Perhaps the default font on Ubuntu is not Ubuntu?

Oh, right - this computer is PopOS which is a flavor of Ubuntu. They definitely could use a different font.

https://github.com/pop-os/gtk-theme#recommendations > Interface: Fira Sans Book 10

Eh, I don't really mind. It just looks a bit strange.

yeah I removed color because it was causing problems for users with dark mode themes

still wanted something to "stand out" ... this seemed like a "standard" unicode char

What's it supposed to be? I'd like to put a picture to the "rightwards finger-post arrow" hjaha

Thanks 🙂

yeah ok. looks simple enough.

cool, thanks, will prepare the PR, this is not urgent

added something that'll make you happy 😉

https://github.com/thheller/shadow-cljs/commit/94c99bd609bf674e24b47964d37c89e57e838414

hallelujah! thanks a lot. so the PR is no longer needed I guess 🙂

btw. this could also solve some old issues with dirac+node.js https://github.com/binaryage/dirac/blob/master/docs/node.md#source-maps

I only added it because it seems to be the only way to make chrome exts load source maps

it'll probably have issues on all other targets

yep, I didn’t find any good solution either

I still think adding that :source-map-prefix could be helpful for some future ad-hoc workarounds in similar cases when I want source maps to be server independently

e.g. via dev-server or elsewhere

we don’t need to document it, it will be just hidden option for desperate people like me who go digging… ;()

as for the other stuff one option is instead of loading the content-script file directly

it just loads a shim that loads the real files via fetch

I tried that, but gave up, fetch part was easy, but I was running into google.provide throwing that namespaces were already declared

that seems to work fine but requires adding https://developer.chrome.com/extensions/manifest/web_accessible_resources which is a bit annoying

but you have better understanding of the whole closure thing, so you probably know how to do it properly

yeah if you try to load the files over the old ones that happens

just don't load them over the old ones and you are fine 😉

nevcermind, I was probably doing some other mistakes, it wasn’t clear to me how to configure the build properly, that the code I’m fetching is a module which depends on what is already present in the shim

that was the main issue, it contained the shim bootstrap code again

and that is why it was complaining

reading that link you posted, I didn’t run into this problem

I used js/fetch to get the files from dev-server running on localhost, and it worked

probably because you were still accessing files from the dev server isntead of the extension

I think it would be acceptable for dev mode, to have dev-server auto-configured

you can chrome.runtime.getURL("out/content-script.js")

and fetch that

or you could tunnel this stuff through shadow-cljs devtools ws connection

I'd rather not

ok, so you found a solution without relaxing manifest rules? via chrome.runtime.getURL?

no , that way requires the manifest config

ah, ok

maybe I'll just make it talk to the shadow-cljs server.

need sleep, might look at it more tomorrow

sure, thanks a lot