Hacker News new | ask | show | jobs
by discarded1023 383 days ago
It's a nice write up but I'm not sure what the contribution is. I would have expected more engagement with Peter Van Roy's work, in particular his and Seif Haridi's epic book CTM [1] where the logic variable got properly unpacked more than 15 years previously. Or at least a citation.

Has anyone looked into how to decouple logic variables from backtracking? i.e., is there a good reason to unbind a variable apart from the Prolog discipline? (Without unbinding we get single-assignment variables where initialisation is decoupled from declaration, which I feel can often be simulated with laziness ala Haskell, but see CTM.)

[1] https://webperso.info.ucl.ac.be/~pvr/book.html
1 comments
nextos 383 days ago
They cite one of the works of Gert Smolka (Smolka and Panangaden, 1985), who was the original creator of Oz. But I agree that Van Roy and Haridi did a lot of research on functional logic programming back in the 1990s that seems to be ignored here.

When I worked on program semantics, I had the impression that PLT communities in the US and EU tend to ignore each other. This also translates to education. CTM is an epic book, and very readable, but I don't think it's well known or used much on the other side of the pond.

link
discarded1023 383 days ago
Yes, I was unaware of Smolka's early work and it does look very nice. (Much of Smolka's work seems underappreciated.) The evaluation in the paper under discussion of this work seems weak given that Smolka's early effort has a very similar set of features to what they want.

I concur about the research silo-ing. However in this case we have Europeans ignoring Europeans, and on that CTM page you will find a grab from a review in the Journal of Functional Programming. Several authors of this paper have published in the JFP, leading me to conclude that the JFP is a write-only journal, like much of CS literature.

link
cryptonector 383 days ago
Could it be that they're not aware of that which they are not referencing?

I notice that they reference Icon. They don't reference jq (well, there's no publications on jq to reference).

link