tldr: everything in pieces on the shop floor, but....
I'm trying towards an implementation of Babashka Pods in Haxe.
Given Haxe is a polyglot language, it makes sense to go for a polyglot interop solution.
There's things like built in async handling, client side code and platform awareness, very much the inverted pyramid that's good for adoption by keeping complexity at the base low.
A Haxe Pod implementation would transpile to around 10 different languages, and could be a base for a repl implementation, certainly of HScript, and or KissLang or some of the other languages and dialects cropping up.
Looking at what I did with Stig this week it probably bears some explanation.
I wanted a structured data format that doesn't rely on reflection, doesn't require the heavyness of the internal runtime representation of tagged unions (enums/gadts) but can be pattern matched on.
The external StigForm is neither here nor there, the internal StigNode is the important bit. There's probably a few minor adjustments so it can sit well in most languages.
I might well use it to build a Bencode output, as it fits perfectly.
The inclusion of Stx, or the dead-code eliminated version of it is for parsing is unavoidable, but the generated loc is only around 5k at present.
It's based on Stephan Le Dorze's mix of Haskell Parsex plus Alex Worth's Phd work on linear time left recursive packrat parsing, and I've been tinkering with it for years.
Work on Go2Hx by PXShadow means there's the opening to cross compile CueLang, meaning that there's an opening for language native type generations from Cue, which is pretty exciting, and Haxe has an extremely compact and well designed internal type representation, and the new completion algorithm is very snappy.
Still work to do though, I'm in the middle of a rebuild atm.
#Haxe #Programming #Babashka #Edn #Pods #CueLang #GoLang #OMeta #Parsex #Interop #DevOps #OMeta
robrich
Jarno 
Adam - K3CAN
Kevin Karhan 
ohmrun
Israel
kurtsh
Big Labie Energy the Podcast
:pressure network