interested in implementing server side of Castra in Ruby. Am I nuts? Is it possible? What do you guys think?

sure, why not?

it was originally implemented in php

i was looking a bit deeper into javelin over the weekend, because i was trying to make the cell graph reloadable. @micha i remember you mentioned something about the lack of weak references and garbage collecting cells. do i understand well, there is no way to reset the formula cell graph by just using boot-reload, while still preserving the stem cell state?

Is there a way to require a .hl file from the boot repl so I can play with functions in the repl?

Sorry if that’s an obvious question, I’m pretty new to boot and hoplon and have been looking through the docs. I’ve tried a few things in the boot repl and haven’t found it via some obvious attempts.

hl files are processed by the boot-hoplon plugin which converts them into .cljs files

which are then compiled by the Clojurescript compiler

you can start a cljs repl which, when connected to some javascript runtime, will allow interacting with the state

i see, i’m trying to connect my JVM to something running in the browser 🙂

yeah, you need a Clojurescript repl

ah, of course, ha

https://github.com/adzerk-oss/boot-cljs-example

ooh, thank you

so, once I’ve got the browser repl hooked up (and it seems to be working already), can I change ns into my .hl ns?

(via in-ns)

hmm, I just did a clojure.tools.namespace/find-namespaces-on-classpath and didn’t find my namespaces from the cljs-repl

I don’t think it works that way

tools.namespace only works for Clojure namespaces

I’m not sure you can in-ns in Clojurescript, however I’ve never tried

can you run the fns that you’ve defn’d in a .hl file?

yeah

oh, so do you have to require a ns?

did you try (require ‘your.ns)?

I have, but get a file not found

due to the way cljs works all your namespaces accessible from the main script will already have been loaded

by the time you’re connected to the repl

I’m pretty sure you can’t load cljs namespaces dynamically unless you eval them

for example, what would the ns be for a file with (page “chapter1-1.html”) at the top

located in src/chapter1_1.cljs.hl

I don’t really know 🙂 you’d have to look at the boot fileset or with the devtools in the browser

I don’t use cljs repl

ha, okay, thanks

well @dm3 thanks for your help, I feel like I’m closer. I better get some sleep now though.

np

@mudphone the cljs namespace name is generated

you can see it in the js console if you put something like (print "hi there") in your .hl file

the devtools should show the namespace, file, and line number in the console

line numbers should be preserved

you should be able to require that namespace

in the cljs repl

@micha thanks

When I add (print “hi there”), and look in the browser console, I see:

hi there  …  repl.cljs:47

oh interesting

do you get a stack trace?

I’ve found this informative: https://github.com/clojure/clojurescript/wiki/The-REPL-and-Evaluation-Environments 🙂

@micha are you sure the require will do anything? The wiki above indicates you need to use load-namespace/`load-file`

well you can always do (print (.-stack (js/Error. "hello")))

in your .hl file

oh, I found this in the browser: hoplon.app_pages._chapter1_1_DOT_html

there we go, that's it

in the goog.require

Thank you. I clearly have a lot to learn.

it does that because you could have a html file named for instance 1.html

which isn't a valid symbol

symbols can't start with a number etc

also you don't want generated namespaces to collide with other ones

so it namespaces them in the hoplon.app-pages package

and prepends a _ to your page name

and munges it

i am not sure anymore about the procedure for loading a namespace in the cljs repl

do people not do this often? i’m new to this so, maybe it’s not a normal thing to do?

i use the live reload instead of a cljs repl usually

i find it's pretty simple to just do like

(cell= (pr some-cell-i-want-to debug))

in a file and when i save it i see the output in the js console

ah, okay, I think I remember seeing that in one of the docs

i suppose one other option is using .cljc files and using the normal repl?

you still need a cljs repl

separate from the clj one

the clojure compiler will ignore cljs parts

oh, I meant to just test a normal fn, not the hoplon bits

i guess that wouldn’t be real useful

some people swear by the cljs repl

I guess it works fine once you develop a workflow

there’s also Dirac and cljs-devtools

Thanks for the tips and assistance. Much appreciated.

@mudphone I think there are real advantages to keeping as much of your app as is reasonable in .cljc files. One is the JVM REPL. As cool and handy nas the CLJS REPL is, it still lacks a number of things that the JVM REPL has. E.g. vars, but also tooling.

@candera for jvm repl is .cljc better than .clj files?

Not better per se.

If I’m doing a bunch of stuff that’s clearly in JVM-land, then I go with .clj rather than a .cljc with a bunch of reader conditionals.

All sort of depends.

But often there’s a bunch of purely algorithmic stuff, and I find that putting that in a CLJC doesn’t cost me anything and lets me use it in my CLJS app but still play with it in my JVM REPL.

ah i see

that makes sense

@candera thank you, for some reason, now that you've said that, I'm wondering if I heard you say that on a podcast :)

🙂

Maybe - I don’t listen to anything I say.

Ha, well, I really enjoyed your show (still do). I think was my favorite, you certainly have a knack.

Is this the best way to check out cljc? https://github.com/clojure/clojurescript/wiki/Using-cljc

(Best or good enough :)

@micha Thanks, I just needed a reality check of whether I was being insane thinking of it. I would love to be able to start converting to hoplon from within a Ruby app, but I would rather use the RPC style than other.

@mudphone Thanks for the kind words. Glad you enjoy the show. That wiki page seems like a decent place to start.

The process is pretty straightforward, though. If you’re in the JVM, and try to load namespace foo.bar, the compiler will first look for foo/bar.clj, then foo/bar.cljc. If you’re in CLJS land, it will first look for foo/bar.cljs, then for foo/bar.cljc.

Yes! I remember seeing that load order in the errors I was generating while blindly searching for my namespace in the repl.

It can make sense to have both a foo/bar.clj and a foo/bar.cljs. I do this, for example, to provide alternate versions of a set of functions for reading files depending on whether I’m running in Node or the JVM. But I can’t think of a reason to have a .cljc and either of the other two. And of course you can use reader conditionals in your .cljc to get different code depending on which context you’re in. That’s handy from time to time, but I find myself staying away from writing too much code like that.