Interestingly, when marshaling integers less than 200, they are converted directly to a corrsponding byte
so
(print (marshal wrapped)) should show the string
but that is the array being marshalled, not the original string
As I recall, it's not tree-shaking the standard library yet, but it's a possibility - is that right?
Functions from the core library will have tree shaking and there is a way to not allocate the standard library at startup - however, there is no guarantee that the core library bytecode will not be included in the final binary.
is there a robust way to get the arity of a function/cfunction?
disasm works but not on cfunctions
that is,
((disasm nil?) :arity)
probably not what you're looking for, but fwiw, for functions that take at least one argument, calling them with zero arguments gives an error message including arity info (in some cases?).
i guess one can use
compile and examine the return value too.
e.g.
(compile '(put 1))
gives:
@{:column 11 :error "<function put> expects at least 3 arguments, got 1" :line 8}
not great for sure
there's this neat thing andrewchambers made: https://github.com/andrewchambers/janet-where-defined to find definition locations of things including c functions. i don't suppose it would have any hints -- it does require debug symbols for c functions though.
Hi all! I'm trying to develop a simple slack bot on janet (basically an http server) and I'm using circlet and ijanet on emacs. Right now I have an insane workflow where every time I change something, I have to interrupt the entire janet process (C-c C-c on emacs) and restart the repl, reevaluate the buffer, then start the server. The experience with vscode is pretty much the same. Is there a better way?
If you want to keep state around, you can implement defonce: https://github.com/janet-lang/janet/discussions/607
I don't know anything about circlet though
ah sorry, i phrased it weirdly. I'm running ijanet on emacs (and yes i have the eval buffer) and i'm trying to use circlet, which is a webserver based on mongoose. It has a
(while true) to keep the server running, and it seems like the repl is blocked while this is happening. Having said this, I realize the solution might simply be to wrap it in a coroutine. That would make it async, and perhaps not block the repl.
and use vars
and manually code some reloading logic
I solved a similar problem with netrepl and ev/call. Relevant bits of code: https://github.com/Saikyun/text-experiment/blob/main/src/main.janet#L503
I also used dofile to get an env and merge it into the root env: https://github.com/Saikyun/text-experiment/blob/main/src/file_handling.janet#L54
@sogaiu:matrix.org has some info about using netrepl from Emacs, but I'm not sure if that code is public
i think the emacs netrepl code i have is here: https://gist.github.com/sogaiu/aa957885d965e4b874fc83c55f873719
@subsetpark: may be you found it already, but iiuc, ijanet refers to this: https://github.com/SerialDev/ijanet-mode
e.g.
(:+ a b)
it seems pretty clear what is going on
so the answer to writing a polymorphic formula would have been to use methods for everything
i guess that would have required methods on things like Number
the syntax doesn't even look that bad imo
i guess it would put you in the situation that to use int/s64 you would be required to use methods
poly/pos? poly/max poly/min ...
and they would be implemented in terms of methods
I think its currently quite error prone in the stdlib to use any of the standard functions with abstract number types
since a bad comparison often silently returns true or false
due to total ordering
im not sure the real value having + call a method on fallback provides
though i think it might not be that bad to do static method lookup for primitive types