I've actually been thinking about this quite a bit. I remember a foray into prolog when I was a younger pup in 2004-6. With the Advent of llms, I think that perhaps we could use llms to extract triples from large corpuses of text and then use that to build our prolog stores or ontologies and work on them. I haven't really experimented much with it but you saying this has reminded me that I should dig that back up again.
Ask HN: What's Prolog like in 2024?
21–30 of 289 posts
Re: Ask HN: What's Prolog like in 2024?
#22Definitive reference: https://www.urbanautomaton.com/blog/2015/08/10/the-pledge-to...
Looks fun :D, i think that if i ask my manager to build something out of Prolog i would probably get stab... i mean fired since most of us work in OOP. I would love to be that insane one asking for that :D.
You can use https://logtalk.org for oop in Prolog, use it on top of SWI and you have bidirectional bridges to Python an Java
https://www.swi-prolog.org/FAQ/Python.md
https://www.swi-prolog.org/pldoc/doc_for?object=section(%27p...
Re: Ask HN: What's Prolog like in 2024?
#23Shameless plug: you should check out my podcast The Search Space for a view of the broader landscape of Prolog and logic programming: https://thesearch.space/
I don't publish episodes often but I have a lot of good interviewees lined up :)
In general, I would advice you to look beyond Prolog and explore Answer Set Programming, the Picat language, and the connections between logic programming and databases (SQL, RDF or otherwise). Not instead of Prolog, but in parallel. Prolog is awesome!
Re: Ask HN: What's Prolog like in 2024?
#24Shameless plug: you should check out my podcast The Search Space for a view of the broader landscape of Prolog and logic programming: https://thesearch.space/ I don't publish episodes often but I have a lot of good interviewees lined up :) In general, I would advice you to look beyond Prolog and explore Answer Set Programming, the Picat language, and the connections between logic programming and databases (SQL, RDF o…
ASP is in another uni course of mine ;). I'll check the podcast, thanks
Re: Ask HN: What's Prolog like in 2024?
#25Not sure about Prolog itself but Datalog really needs to overtake SQL, it's just so much better. Related areas like constraint programming are still very relevant.
Could you explain more or point out some interesting references? I'm currently trying to understand how Datalog compares to SQL and, potentially GraphDBs
Re: Ask HN: What's Prolog like in 2024?
#26Prolog itself is still developed and used in various settings (mostly swi-prolog?), but other languages and logic engines solve domain specific but similar problems better (rule engines, formal proof verifiers, etc). For exploratory work it can be useful. I have tried to use it in combination will LLMs unsuccessfully, partly because the domain was not specific enough. Otherwise you need a lot of real world knowledge…
Any answer here is a good one since the question is soooo unspecific :D. My professor is a staunch advocate for RDF/OWL, inference engines and stuff like that (hence why i also mentioned ontologies :D). The thing is that i think that the language itself has so much untapped potential and the world that i dived into with my studies is so vast, so full of stuff that it left me kind of dazed to be fair! I got some paper…
You are definitely on to something here. OOP has some common roots with formal ontologies and knowledge representation (not so much the programming languages, but object oriented modeling). OO fails at this for various reasons, whereas logic is tailored for this specific purpose. Check out ErgoAI (formerly Flora-2), it's the most advanced Prolog flavor for representing and reasoning over knowledge. https://github.com/ErgoAI
Re: Ask HN: What's Prolog like in 2024?
#27My professor swapped Prolog out for Rust at the last minute. I don't know whether he did us a disservice or a favor.
Re: Ask HN: What's Prolog like in 2024?
#28Prolog has reached an exciting new milestone with Scryer prolog. It is the first highly performant open source iso-compliant Prolog. I would check out Markus Triska's work to have your mind blown: https://www.metalevel.at/prolog https://youtube.com/@thepowerofprolog
Do you have any papers comparing Scryer with other prolog systems (like SWI-prolog or SICStus prolog) performance-wise ?
There are some benchmarks here of SWI Prolog's benchmark suite on diffrent Prolog systems by Jan Wielemaker the SWI Prolog author:
https://swi-prolog.discourse.group/t/porting-the-swi-prolog-...
He finds Scryer performs worse, which he does comment on, he also explains some tradeoffs and historic choices in SWI's design which affects its performance. I think I have seen the author of Scryer saying that's not surprising and Scryer is still building up core functionality where SWI has had 30+ years to optimise, but I don't remember where I read that.
SWI has a document explaining some strengths and weaknesses regarding performance: https://www.swi-prolog.org/pldoc/man?section=swiorother
Edit: some discussion on Scryer previously on HN: https://news.ycombinator.com/item?id=28966133
Re: Ask HN: What's Prolog like in 2024?
#29No idea, but it might be worth looking into Mercury and {mini,micro}Kanren/core.logic as more practical iterations on it (either by adding things to Prolog or extracting the interesting to stuff to use in more general purpose languages).
At the end of the day "practical" means library support and community knowledge, by which measure Prolog and more specifically SWI and Sicstus are far more practical than any of the other logic languages or implementation options
Re: Ask HN: What's Prolog like in 2024?
#30Prolog itself is still developed and used in various settings (mostly swi-prolog?), but other languages and logic engines solve domain specific but similar problems better (rule engines, formal proof verifiers, etc). For exploratory work it can be useful. I have tried to use it in combination will LLMs unsuccessfully, partly because the domain was not specific enough. Otherwise you need a lot of real world knowledge…
Hi! Are you Jim Hendler (or related to him), my Reagan-era AI professor from UMD?
https://en.wikipedia.org/wiki/James_Hendler
My Prolog programming assignment #4, a Prolog "nehcihsahA" detector (maternal uncle: a mother's brother, or any equivalent relative) seemed designed to make me hate Prolog with a passion, involving bending over backwards by defining ridiculous predicates like siblish, sibloid, relatoid, sistoid, brothoid, mothoid, and fathoid.
https://www.donhopkins.com/home/code/nehcihsaha.prolog.txt
I much more enjoyed the OPS-5 programming assignment #6, for which I made a worm simulation that hacked into Ollie North's Intimus-007s ("the ace of security paper shredders") in the White House basement, via Professor Hendler's Sun workstation dormouse, rms's account with password rms on prep, and Casper Weinberger's account on UMD's Vax 11/780 mimsy and NSA's PDP-11/70 tycho, connected via the NSA's MILNET IMP 57 at Fort Mead, then posted Ollie North's secret diary and notes it found in the paper shredder to talk.rumors via the UCB-Vax usenet gateway.
https://www.donhopkins.com/home/code/crack-ollie.ops5.txt
https://news.ycombinator.com/item?id=18376750
>At the University of Maryland, our network access was through the NSA's "secret" MILNET IMP 57 at Fort Mead. It was pretty obvious that UMD got their network access via NSA, because mimsy.umd.edu had a similar "*.57" IP address as dockmaster, tycho and coins. [...]