it would be nice to not have to do method calls :) ideally when building / not repling, everything would just act as regular function calls
But if you do this
(table/getproto (fiber/getenv (fiber/current)))you get
@{os/clock @{:value <cfunction os/clock> :doc "(os/clock)\n\nReturn the number of seconds since some fixed point in time. The clock is guaranteed to be non decreasing in real time."} net/write @{:value <cfunction net/write> :doc "(net/write stream data &opt timeout)\n\nWrite data to a stream, suspending the current fiber until the write completes. Takes an optional timeout in seconds, after which will return nil. Returns nil, or raises an error if the write failed."} bytes? @{:source-map ("boot.janet" 100 1) :value <function bytes?> :doc "(bytes? x)\n\nCheck if x is a string, symbol, keyword, or buffer."} ...}
Looking at your example code again, I realise that my problem is slightly different. Yours looks to me like a variation of the 'reloading namespaces' kind of problem (do you have to do it?). I think what I'm thinking about is that if you're in one file in an editor that uses the netrepl and you send the content from that buffer to the netrepl server it's evaluated in the current environment. If you then send the content from a different buffer to the netrepl server, it's not evaluated in a different environment (at least it isn't at the moment).
Clojure REPLs (and possibly others) get around this problem by having the REPL environment be always in a particular namespace and allowing the user to manually change the namespace. But Janet's REPL doesn't really work like that. There is sort of just the REPL environment and everything you type in it goes into that environment.
Maybe I got it all wrong, but why not just
(put modules/cache "my-lib" nil) @Saikyun ?
but again I can be totally not understanding at all :-)
if I had to wager a guess, the
modules/cache-part will solve a similar problem as import :fresh true, but I wonder if it will solve the "thrown away names" problem
I'll try it later and get back to you :)
thanks for the info
oh I am going to speculate
(import :fresh true) is doing that module/cache thing :-)
so no help probably
so thanks again
@pyrmont as you say, those are two different problems :) for me, it's all about being able to reload "namespaces". in my case, I have a render loop in
main.janet, and it depends on various files. in those files I want to make changes and have them come through to the render loop. wrt: evaluating in different environments -- an environment and/or fiber per netrepl connection might make sense? :) not sure what problem you want to solve
can you please share yours (as you seem disapointed 😀
I'm trying to test various things with a stream to a netrepl connection from an ordinary REPL. I have two functions,
recv and send which read from and write to the stream respectively. The problem is that if I read from the stream when there's nothing there, it will block.
in my mental model I tied
ev and net as tight as my old mind allowes
I guess those functions aren't exceptionally complicated so I could just tweak them.
don't know if it will help you, but this is how I manage to get netrepl + loop to work together: https://github.com/Saikyun/text-experiment/blob/main/src/main.janet#L217-L235
the
ev/sleep call makes it so that the netrepl gets a chance to do its thing :)
when thinking about it, a generator might have been cleaner, haha. anyway :)
I'm running the netrepl server in a separate process so it's OK for
recv to block, just not indefinitely :)
I was trying to get Conjure to work with it but I am not having much luck and am trying to see if something's broken in the implementation (I don't think it is but wanted to double check).
dogfooding
More like bearpoking :P
no danger I would say