Exotic Programming Ideas (opens in new tab)

I feel like its fair to call logic programming, including prolog and datalog, exotic. They've fallen out of the mainstream (especially compared to lisp which was a big competitor to it at the time) and are quite different to program in than procedural languages.

My mental model of prolog is to imagine you are constructing a graph, where some of the nodes have side effects, and then tell the compiler to do a depth-first-search, executing any side effects it comes across.

Do you have an example of a truly exotic programming concept? I'm really interested in weird PL stuff.

The classic Kernighan and Pike book, The Unix Programming Environment[1], has a good example of advanced make usage (cascading / recursive makes, multiple (and implied) rules, sources and targets, etc.), along the lines of your example, in the chapter at the end of the book on hoc[2], the higher order calculator. They build multiple, incrementally more powerful versions of hoc in that chapter, as a case study to tie together many key topics in the book.

[1] https://en.m.wikipedia.org/wiki/The_Unix_Programming_Environ...

[2] https://en.m.wikipedia.org/wiki/Hoc_(programming_language)

Two fun blogs posts in this vein

https://bentnib.org/posts/2015-04-17-propositions-as-filenam...

http://conal.net/blog/posts/the-c-language-is-purely-functio...

I never thought about make as a logic program.

But considering this seems to explain a lot. Like SQL, make is an integral part of the programming world ... that most people want as little as possible to do with and which some people violently hate and a few people enjoy playing with.

And I think this shows language that just tell the computer "what you want" and not how to get that something tend to not scale with the length of the command or the amount of data processed. SQL was originally created to be something like a human-friendly, natural language interface, if you can believe it. And it's sort of friendly if you only use one or two command-words but after that it gets extremely frustrating.

So the connection is interesting.

Make uses dependency graph to track the make rules. Topological Sort can turn however deep dependent rule network into a simple linear ordered list of steps to perform. The matching files are then fed through this list one step at a time and run the matching rules against the files. Repeat the process for newly generated files.

I think the term you're looking for is constraint propagation.

That's a nice programming paradigm, unfortunately a bit obscure now.

Related: make obviously makes a big dependency graph to keep track of this stuff. Is there a good way to make it show this? How do you debug these kinds of constructions?

I skimmed at first and thought the author was going to talk about the implementation of dependency resolution in apt-get.

Bitcoin as cult, financial fraud, criminal scam, and religion. I like this guy.

What I didn’t see was any tirade against people? - you don’t tirade against people being taken advantage of by scammers - unless those people are the cult leaders?

> “New religious movements like the cryptocult provide a psychological and philosophical framework that provides sense-making for a world that seems hostile and out of their control. The crypto movement fits all the textbook criteria, it provides a mechanism for determining an in-crowd and an out-crowd (no-coiners vs bitcoiners). It gives a framework for assessing the virtue of other followers based on their faith (HODLing) in the cause. It offers simple answers to complex issues in economics and monetary policy. It gives a linguistic framework of “thought-terminating clichés” and acronyms to quell dissent. It gives a mechanism of social control in which one can acquire influence and status in exchange proselytizing and onboarding more followers to buy tokens. It makes miraculous promises of wealth, not derived from effort but from faith. It presents an eschatological narrative of retributive justice about the end-times of the global financial system, in which the true believers will be reborn with a new life in an anarcho-capitalist utopia. And most importantly, it gives people a sense of a community, hope and belonging which is a powerful force that can be exploited by charismatic leaders.”

Can't edit my post, but for context Stephen has since edited his article. My post was related to this version -https://web.archive.org/web/20200730131647/https://www.steph... - which specifically mentioned companies I worked with and respect.

This is great. Thanks for sharing. I echo the author's sentiments and I think you ought to choose better colleagues.

It's best to seperate the human from their technical work.

I do technical work and unfortunately can't contain my Politics on social media (due to their psychology tricks). I know both Republicans and Democrats will dismiss my math because I don't belong to their side.

Yeah, the articles title is: “Exotic Programming Ideas: Part 4 (Datalog)”. All the previous times this series has made the front page, it has had the part in the title.

Yeah, I think that minikanren is a more useful language for embedding than datalog since it is more expressive than datalog and has a more straight forward execution model. The functional style of minikanren makes it easier to create things like parsers easily.

I'm searching for an intro to miniKanren that's also an intro to logic programming. Ive never used datalog, clojure, or others. The main miniKanren website assumes preexisting Scheme / logic programming knowledge

Any suggestion of where to start?

I hope he will cover some other exotic ideas like program inversion which I find really interesting, in the same area there is also specialization, though that is maybe a bit too similar to term rewriting.

The quine generator example is indeed very impressive...

Ed Kmett did a live coding series on his attempt port kanren to Haskell. Hope guanxi makes some progress soon.

Those clojure implementations are not datalog. They are an attempt to get datalog-like functionality in a lisp syntax. At best they translate a subset of datalog concepts into a language that is more general purpose in nature.

Gerrit used prolog to determine when to merge patches if i recall. I had to learn prolog in university once upon a time, but gerrit is the only time i have ever encountered prolog in the wild, and it only uses it i one small spot.

The immutable database Datomic (from Rich Hickey, inventor of Clojure) is all based on datalog, and is fairly popular in the nichey world of Clojure development; there are several other similar datalog-based systems outside of Clojure as well.

Various marketing case study blurbs are here if that's your thing: https://www.datomic.com/nubanks-story.html

SPARQL isn't too far off! and Wikidata provides a really fun query service for it. https://query.wikidata.org/

Wow, I'm not sure why this was unpopular, does someone want to fill me in?