Fork me on GitHub
#babashka
<
2020-01-20
>
sogaiu15:01:17

nice about the changes to support different ns per socket :thumbsup: (that's not something that shadow-cljs does unfortunately, making multiple connections to a process less useful -- actually that's the last i checked...may be things are different now) also appreciate the :reload support

sogaiu15:01:55

all tests pass here for: 55238b52acc086bb08e91161a3bfdba32171e55f

borkdude15:01:14

good to hear. thanks to @mauricio.szabo for prompting me about some issues with the socket repl

sogaiu15:01:53

@mauricio.szabo have you asked thheller about supporting a different ns per socket?

mauricio.szabo15:01:34

@sogaiu not really, the issues on bb were to support changing ns on a socket repl. Commands like ns and in-ns were failing when connected via socket.

borkdude15:01:30

I discovered the issue with per-connection ns after fixing that

borkdude15:01:57

I kept track of the current namespace in two places which went out of sync, but now it's only one place.

sogaiu16:01:24

@mauricio.szabo ah ok, thanks. what i found is that for 2 sockets that are each in a cljs repl, they end up always being the same ns. if you in-ns in one, the other changes too.

sogaiu16:01:01

cool that things have improved in bb :)

borkdude16:01:22

I'm guessing but for cljs it might be related to dynamic vars not being really dynamic vars, since there are no threads, they cannot have a thread-local binding

sogaiu16:01:13

may be - though iiuc in-ns is "emulated" in the repl, and the repl itself is implemented in jvm clojure.

borkdude16:01:09

fwiw planck seems to do it correctly, although it's mostly JS + a bit of C

sogaiu17:01:32

ah i didn't know that -- nice!

borkdude17:01:37

Also tested lumo which also does it correctly, runs on node

sogaiu17:01:10

just verified for planck -- haven't installed lumo yet

borkdude17:01:39

if you have node, it's just npm install -g lumo-cljs

sogaiu17:01:15

thanks -- i'm building from source.

sogaiu17:01:58

verified -- lumo seems to work well in this regard. thanks for testing and mentioning these two :thumbsup: