Why Learning Prolog Can Make You a Better Programmer | Hacker News Reader

Why Learning Prolog Can Make You a Better Programmer

coliveira.net

53 pointsgnosis30 comments

Prolog's ideas are alive and kicking in numerous interesting forms:

  * Constraint Handling Rules, http://dtai.cs.kuleuven.be/CHR/
  * Gecode http://www.gecode.org/
  * Monadic Constraint Programming Framework (Haskell)
    http://people.cs.kuleuven.be/~tom.schrijvers/MCP/
  * Google Operations Research http://code.google.com/p/or-tools/
  * Mozart/Oz http://www.mozart-oz.org/
  * Datalog 2.0 http://datalog20.org/

Interesting timing for me -- I've been working through "The Art of Prolog" (MIT Press, 1994) recently -- I'd done some Prolog in school, but never beyond the ``here are a few weeks in a programming languages course'', and the logic programming discussion in both SICP and ``The Reasoned Schemer'' had made me want to go back to the source.

I'm about 2/3 of the way through the book, and I highly recommend it to anyone looking to learn the language in a more or less rigorous way. It won't make you a production Prolog programmer (any more than SICP will make you a production Schemer), but it starts with a firm grounding in the theory of logic programming, and works that up to a good grounding in Prolog step by step, before spending the latter half of the book working through idiomatic Prolog solutions to a bunch of standard (once-standard?) problems (a shell, an interpreter, a compiler), as well as problems more in line with Prolog's traditional uses (ELIZA, an expert system).

The book is here:

    http://www.amazon.com/Art-Prolog-Second-Programming-Techniques/dp/0262193388

``The Craft of Prolog'', which this post mentions is in the same series, and provides somewhat of a more pragmatic view of the language. I may get to that next. Meanwhile, ``The Reasoned Schemer'' provides the entry of ``the Little Schemer'' series into the field of logic programming, using the MiniKanren logic programming system for Scheme.

I second The Reasoned Schemer as a fantastic introduction to logic programming. You can get miniKanren and its variants here - http://code.google.com/p/iucs-relational-research/. They'll run in pretty much any R5RS Scheme.

If you're a Clojurian you can give my bells & whistles implementation a spin: https://github.com/swannodette/logos. It has good performance, tabling, pattern-matching, and disequality constraints.

_The Craft of Prolog_ has a lot of good bits, but is written in response to something less readily available, so its organization seems a bit weird.

I also recommend the hell out of _The Art of Prolog_. Have done so here, many times. Also: Prolog systems with constraint extensions are MUCH more powerful. SWI Prolog (http://www.swi-prolog.org/) and GNU Prolog (http://www.gprolog.org/) are both good.

Art of prolog makes an excellent first book, and the Craft complements it well with seasoned advice on elegant and efficient prolog.

Unimpressed with article. For example, it says prolog only returns true and false. Prolog only returns "i proved it is true" and "I couldn't prove true", which they mislabel as false - so false is not "not true". If you want to focus on features that expand your understanding of what a programming language is, that is surely a big one - false isn't !true? Wrap your brain around that and try to program anything that is still correct, it breaks everything you know about what you can do with a computer.

I disagree with seeing prolog classified as functional - my experience with functional programming in lisp had a very very different view of side effects than my experience with declarative programming in prolog.

Sure, learn prolog and you'll learn more about programming, but that's true of any programming language. I'm not convinced the benefits of learning prolog exceed the benefits of spending the same amount of time on something else. The main thing I learned in a prolog class was the huge danger of relying heavily on facts without also considering confidence in those facts. I can learn the exact same lesson if the declarative language I choose is SQL, with the added benefit that I might get to use the SQL someday.

I've always liked prolog, and seeing prolog love on HN makes ?- happy(cydonian_monk) return true. My first exposure to the language, as I'm sure was the case for many of you, was in an Intro to AI course. Though I've never found the opportunity to use the language professionally, I do on occasion drop it into personal projects.

That said, I've always considered prolog to be the language that most threatens our sanity. Perhaps it's because the Graduate Assistant that taught my Intro to AI course was a bit cracked (he tied his shoelaces around his legs once in class because they "offended him"). Or maybe because I haven't been entirely sane since that class....

"Prolog is a pure functional language ... so there are no variables to update?—?only method calls."

It's been a while (some decades) since I wrote my last prolog program but this seems to be rather wrong to me. First of all, there is also the paradigm of logic programming and prolog is it's best known child. I don't think you can equate logic programming to functional programming. Secondly, there are no method calls as such. IIRC a prolog interpreter basically does a depth-first search on a graph made up of assertions. The very point of prolog is though that you don't tell the interpreter "call this method with these arguments and then do that" but rather you tell it "show me valid solutions that fit this template assertion, do however you think is best". At least that's what simple interpreters did. I don't know what more sophisticated compilers like gnu prolog made out of it. Please somebody correct me if I'm wrong. It's possible that my memory fails me.

not that depth first search isn't a pretty good analogy for what happens, but I believe it's SLDNF resolution, and I'm not 100% sure the exact algorithm is mandated by the language, though it might be for cuts.

prolog is actually impure logic language, and given the general lack of HOF i'd be reticent in labelling it functional.

it's a declarative language though, and the model you have 'here is a query, what solutions exist' is reasonably accurate

There is a subset called "pure Prolog", which is restricted to Horn clauses. But regular Prolog (with extralogical features) is not functional, and definitely not pure. For instance, the proof_me predicate in the following program will be true, only once:

  :- dynamic bla/0.

  proof_me :-
    \+ bla,
    assert(bla).

I don't know prolog but it seems to operate on relations and not functions. So my guess it that you are right and it is not a functional language.

Agreed. Back in the 1980s I did my Masters' project in Prolog (transistor sizing, part of a "silicon compiler") and it gave me a lot of new tools for thinking about programming in general.

The same's true for other langauges that offer different ways of thinking of about world: the LISP family, Haskell, Eiffel, Forth, awk, ...

Learning Prolog sounds worthwhile, but time spent learning Prolog is time not learning something else. Why not Erlang or something obscure I've never heard of like Fril?

Also, I'm getting fatigued by authoritative talk about computer science curricula in the same "kids these days" manner as someone convinced the world is falling apart while actual violent crime or teen pregnancy statistics drop. I'd expect claims like Oliveira's to be supported by some kind of data we can be privy to instead of vague anecdotal impressions.

"Schools are more interested in teaching marketable languages." Ok. All schools, most schools, state schools? Where's justification for "It seems like that they are looking for ways of getting people through the program as quick as possible"? Why does it seem that way?

prolog is an interesting take on programming and by comparison, the style in which you write code is still significantly different to approaches in other languages.

sadly, most people when taught barely get past the 'parents and grandfathers' stage of programming.

why not something else? sure! erlang is actually pretty neat (otp is golden) and was originally implemented in prolog. why not learn a term re-writing language? what about snobol?

why prolog? well,

prolog for me has just been one of the most fun and interesting experiences I have had while programming. good prolog code is elegant in a way i've never seen in other languages and it's a big rush when you crack the puzzle of expressing your problem.

prolog was also used in ibm's watson and shipped as part of nt too :-)

I've always wished someone would do a JVM-version of Prolog, like Clojure (Jalog?) where the prolog databases can, on demand, be linked directly to MySQL databases and/or a networked in-memory cache of some sort. Do that and don't charge an obscene amount and I can promise you that you'll have at least one paying customer. :-)

I wrote a library for Clojure called Cascalog. It's a variant of Datalog that executes using Hadoop. It's not Prolog, but you may find it interesting.

https://github.com/nathanmarz/cascalog

As much as I agree with the title I find the contents of the article misleading.

It is almost as if prolog didn't make him a better programmer, given the idea that prolog 'returns values' or uses 'method calls' is 'not even wrong'.

Agreed. Though i'd like to have someone name me a language, adding which to one's skillset would make one a worse programmer.

Intercal?

You can waist time; which you could use for other things, making you a worse programmer than if you would have been if you picked Prolog instead :P

Seriously, our attention span, time on this earth and even ability to learn is limited ;)

It is practically impossible to teach good programming to students that have had a prior exposure to BASIC: as potential programmers they are mentally mutilated beyond hope of regeneration.

The use of COBOL cripples the mind; its teaching should, therefore, be regarded as a criminal offence.

-- Dijkstra

http://www.cs.utexas.edu/users/EWD/transcriptions/EWD04xx/EW...

I guess I would recommend perhaps LISP, FORTRAN and APL if you wanted to pick old time languages. More recently you might want to take a look at Ruby and Haskell, to pick a functional language, and COBOL, which has some interesting ideas in it you won't find in some of the other languages.

-- Guy Steele

Notice that he lists COBOL as a language to learn. I've always found Dijkstra to be a little over the top (often the only way to get people to listen), and Steele more pragmatic.

http://www.infoq.com/interviews/Programming-Languages-Guy-St...

Dijkstra was provably completely wrong about this: pretty much all the most talented programmers I have ever talked to or read about started with BASIC.

The starting language matters very little, the important part is to remain curious and to keep learning.

From the article.

"In Computer Science programs, there is a tradition of having a course of two that represent the watershed between people that can grok CS concepts and the ones that can’t. "

Would those HNers who've been through a masters or PhD in CS help an uneducated brother hacker and list any such "watershed" courses they know of? Thanks in advance.

EDIT: links to course home pages would be highly appreciated, plus any reminiscences of how tough these courses were/felt like.

In my undergrad CS program, it was Operating Systems. The course required us to move from writing programs to solve well constrained problems to addressing issues of memory management, file systems, command line interfaces, etc.

This was the same in my curriculum, and I'd throw in Compilers and CPU Architecture & Design as well.

There's nothing like building a simple OS, compiler, and pipelined CPU to transform "the computer" from a magical black box into a supremely powerful, interesting tool.

I was going to say AI, but then I remembered my AI course had us do a few small projects in Prolog...

Really, though, I'd say AI. Mainly because there is often no a "right" answer, just the "best" one given the inputs and problem space. It gets you thinking in a more non-deterministic way, if that makes any sense.

I recently completed my master's in Computer Engineering. I found Computer Architecture and the assembly & C courses to be very enlightening. In comp arch, you design data flows, learn how memory and cache work, etc. In assembly and C, you then interact with that same hardware in a pretty direct fashion. Great insights into how computers work.

The Haskell modules were the real deal for me, was my first intro to that programming style, and I subsequently carried over the functional style to python, ruby and even php!

Reply on news.ycombinator.com