11:41 <arigato> a cool "core language" in functional logic programming, which may (maybe) be better than the alternatives so far like Oz:

11:41 <arigato> https://simon.peytonjones.org/assets/pdfs/verse-conf.pdf

11:47 <arigato> so far they don't have an actual evaluation strategy, which makes it all a bit pointless imho

11:48 <cfbolz> arigato: right, and indeed the existing functional-logic languages like curry indeed had similar questiosn

11:48 <cfbolz> and various curry implementations had different strategies, iirc

11:59 <arigato> ah, the paper I link to contains a proof of confluence, which I think means that the strategy "pick a random rule next and hope it does not diverge" would be valid

12:01 <arigato> yes, "If R is confluent then every term reduces to at most one normal form."

13:08 <fijal> is functional programming easy to reason about?

13:19 <jean-paul[m]> some forms of pure functional programming are eas_ier_ to reason about than other programming techniques

13:26 <jean-paul[m]> (where "easier" really means there are more tools available for doing so, as well as fewer pieces to keep track of)

13:31 <arigato> fijal: as far as I can see, it's pushing functional-logic programming in an interesting direction

13:32 <arigato> they define the language with a set of rewrite rules, which is new for functional-logic programming

13:33 <arigato> and that means it would be possible to reason about what the computer does step by step, instead of "I write my program that somehow describes what I'm looking for, but I have no reasonable way of knowing if the computer will manage to find the answer"

13:36 <arigato> (there are still several rules that apply in a given situation, and "convergence" means that you get the same answer whenever you do, but they still have to provide an evaluation strategy, otherwise applying the wrong rule might never terminate

13:36 <arigato> )

13:40 <fijal> right, so "easier"

13:42 <arigato> just saying, I'm just not sure jean-paul[m] was answering to that paper at all. your question was about functional languages, this paper is about a specific functional-logic (!= functional) language

13:43 <fijal> is that something more than "lambda calculus with logic variables"?

13:44 <arigato> it's *something* that includes both lambda calculus and logic variables, which are not immediately compatible with each other

13:50 <arigato> also it is "easier" to reason about only in the sense that functional-logic programming is typically impossible to. there is no way I'd say it is easier than python

13:52 <arigato> (and as far as I can tell, the research group is hired by Epic Games and gives their language a name with a link to the Metaverse, but this part is guaranteed 100% nonsense)

14:19 <jean-paul[m]> Indeed I have not read the paper.

18:33 <fijal> is this "you can do whatever you want as long as you name it something with the metaverse"?

18:36 <cfbolz> i think the "verse" name for tim sweeney's lang ideas is much older than current metaverse hype. I heard the theory that it's a pun: "an epic [poem]" is written in "verse"

18:43 <fijal> heh, so not even