Why did Prolog lose steam? (2010)

https://www.kmjn.org/notes/prolog_lost_steam.html

by yeesian • 3 years ago

152 176 3 years ago

pubby• 3 years ago

If I was a company in the 80's or 90's, I'd use spreadsheets if I need declarative data stuff, and I'd use a boring blub language like BASIC for everything else. Few business problems need the type of logical inference Prolog provides.

I do like Prolog though.

contingencies3• 3 years ago

It seems a good place to note that BNF already existed since 1959. https://www.softwarepreservation.org/projects/ALGOL/paper/Ba...

BNF provided a formal grammar allowing the declaration of precise and iterative constructs without additional complexity of algorithmic content. Original spec @ https://www.softwarepreservation.org/projects/ALGOL/paper/Ba...

BNF was created by Naur, the guy who invented FORTRAN, and was critical to both many early RFCs (the most important standards that facilitated the growth of the internet) and man pages in Unix. The syntax:

  executable <required-arg> [optional-arg]

... is a legacy of BNF. https://en.wikipedia.org/wiki/Backus%E2%80%93Naur_form

amoss2• 3 years ago

What connection are you drawing between the two? BNF is used to define language grammars. Prolog is written using Horn clauses. There is not an obvious connection.

jonsen• 3 years ago

"BNF was created by Naur," and "the guy who invented FORTRAN"

sundarurfriend• 3 years ago

I'm always dubious about such answers: "if I was in X camp, I'd use A; and if in Y camp, I'd use B; so C has no reason to exist". There's always room between X and Y, sometimes quite a lot. There's always people that have been shoehorning their problems to fit A or B, and would be happy to adopt C if it didn't have other major issues.

at_a_remove• 3 years ago

I have a Prolog book, somewhere.

Prolog feels like a really useful feature was hauled out of some other programming language and made to stand on its own. Like if I pulled regular expressions out of Perl and tried to make them their own language.

cheaprentalyeti• 3 years ago

I highly suggest the wikipedia entry [0] on regular expressions and their history; the concept goes back to at least a couple decades before Perl, which incorporated them via a c library written by Henry Spencer a couple years before Perl even existed.

[0] https://en.wikipedia.org/wiki/Regular_expression

jjtheblunt2• 3 years ago

they are their own language class...the regular languages, at least in pure form, not the pcre extended regular expressions (which would be a different language class).

https://a.co/d/a5FGZRQ

galaxyLogic2• 3 years ago

Good point. Most programming languages have RegExp support. But very few (?) have Prolog support. That might be useful in some or many cases, like ...

bandrami• 3 years ago

The pattern matching capability in the ML family is arguably an embedded Prolog system.

mechanicum• 3 years ago

It’s not Prolog, but the related miniKanren has been widely implemented.

galaxyLogic2• 3 years ago

I understand that Prolog is good for creating parsers, right? Then could it to a large extent be a replacement for RegExps, providing more powerful ways to do string-manipulation.

User232• 3 years ago

Embedding prolog in Common Lisp was pretty normal during the first wave of AI, as I understand it.

There’s always been an odd irrational hostility to Lisp, so I wouldn’t be surprised if Prolog feels as it does to you exactly for the reason you suspect.

kryptiskt• 3 years ago

Awk is that language. Perl was created as a more capable replacement for that, Awk++ if you will.

rjh292• 3 years ago

Yeah, imagine if Python had a 'declare these two things are equal' operator and let you just randomly summon up computations declaratively, but with all of the python standard library and data structures. That would be more approachable than Prolog imo.

mst3• 3 years ago

Take https://p3rl.org/Language::Prolog::Yaswi and steal the ideas into python then, you can probably at least get close to sqlalchemy style in terms of surface syntax at the python level.

(the perl community has implemented and released many strange and wonderful things, often years before other languages' communities noticed they were there (because we're weird), and while it's entirely reasonable to dislike perl as a language it's well worth looking at the things we've perpetrated over the years and stealing the parts you like for your language of choice :)

mst3• 3 years ago

Funnily enough, perl has both a native prolog implementation AI::Prolog (which is not amazingly fast but serviceable for simple cases) and Language::Prolog::Yaswi, which is bindings to SWI prolog with a perl-native DSL if you want it.

eslaught• 3 years ago

If you're trying to solve boolean satisfiability, you want a SAT solver, not Prolog.

If you're trying to solve more general problems, you probably want an SMT solver or other more specific solver (e.g., ILP), not Prolog.

If you're trying to verify software, you want a theorem prover, not Prolog.

If you want database-ish stuff you want Datalog (or something like it), not Prolog.

Prolog falls into the uncanny valley of being too general of a DSL for the sorts of optimizations that more narrow solvers enjoy, while being too narrow of a DSL for most general-purpose programming. I think because it was an early DSL, people got caught up in the idea of it being more "high-level" than other languages [1], but the DSL nature of it wasn't fully appreciated at the time.

[1]: For fun, take a read through https://en.wikipedia.org/wiki/Fifth-generation_programming_l... , which was a whole thing in the 1980s.

jjtheblunt2• 3 years ago

above is a fantastic lucid comment (not kidding)

usgroup9• 3 years ago

I don’t think those are the problems since all sorts of solvers ship with prolog and it’s still a better way to write down a problem.

I think it’s biggest headwind is that you have to be significantly more intelligent than the average developer to think in Prolog.

Prolog requires you to think about code as the implication of constraints — you just have to keep a lot more in your head. Prolog also requires you to understand the execution model before you can do much of anything.

nerdponx• 3 years ago

What about Prolog as an embedded constraint-solving engine inside an application written in some other host language?

discarded10239• 3 years ago

Prolog also requires you to understand the execution model before you can do much of anything.

This is true of most (all?) languages. Prolog's control flow is ultimately not too different from the usual stack-based semantics people are used to from imperative languages like C (if you're not too ambitious with the backtracking; if you are, best of luck!). The tricky part is perhaps understanding the dataflow that arises from unification. That can be arbitrarily complex but it seems people tend to go no further than difference structures, which are easy to comprehend in terms of pointers. (Yes yes I know, but logic variables are essentially just single-assignment pointers.)

The thing I find deeply weird about the logic programming literature is that if you want to reason about your program you're told to use things like Hoare logic [1] or intended interpretations [2] and not fundamental logical concepts like implementations implying specifications (ala Lamport and many others who grappled with refinement).

[1] see Krzysztof Apt's papers

[2] see e.g. https://people.eng.unimelb.edu.au/lee/papers/dd.html

rurban• 3 years ago

I think it’s biggest headwind is that you have to be significantly more intelligent than the average developer to think in Prolog.

This is arguably wrong. Thinking in Prolog is much easier than knowing how to solve a problem beforehand. In Prolog you just state the facts and known rules, but you have no clear idea what would be the best way to solve it. Prolog solves it for you then, if possible. No intelligence needed. A "dumb" friend of mine on the 80ies only used Prolog for his work and was very happy with it. Lisp, Pascal or C was too hard for him.

The problem with Prolog is mostly that you'll through too hard problems at it, mostly with exponential complexity. Tiny changes have huge influences then, it either works or not. Also the tabling and SAT/SMT integration sucks.

JimmyRuska• 3 years ago

Prolog solvers don't necessarily optimize like gurobi, cplex or OR-tools would. Yes they can give you a solution that satisfies the constraints, and you can query for more possible answers, but it doesn't really optimize for the objective function in the same way.

logicmoo4• 3 years ago

I think it’s biggest headwind is that you have to be significantly more intelligent than the average developer to think in Prolog.

This is very true!

User232• 3 years ago

The Norvig AI book really makes its nature as a DSL clear.

otabdeveloper4• 3 years ago

Solvers are integrated into Prolog nowadays, like the other commenter said.

The real problem is that Prolog doesn't play nice with relational databases. (Too bad the people behind the short-lived NoSQL fad didn't know about Prolog, though.)

baq• 3 years ago

Smuggle a Prolog in YAML somewhere, call it YAMLogic or anything else really which can’t be directly traced to it and watch the kids be amazed at what you can make computers do.

Only about 33% serious here, but… still a bit serious.

maweki3• 3 years ago

Prolog being tied to this very specific execution model of SLD resolution is somewhat root to the problems you have described. The promise of declerative programming in general, and logic and constraint programming specifically is, that programs are easier to reason about because the execution model is abstracted away and there's a relatively simple and mathematically pure definition of the model(s) of a program.

But again and again one uses extralogical constructs in Prolog programming.

Case in point: basically all your mentioned alternatives (Datalog, SMT, I'd add ASP) have implementations that use the Prolog syntax or something close to it with very similar semantics. But not being tied to that specific execution model can give you (better) answers or more complete answers, termination guarantees, or other things we might need.

usgroup9• 3 years ago

I don’t think is true because ASP is not more popular than Prolog but it is essential (or at least much moreso) declarative.

I think Prologs big strength amongst others is the meta interpreter and higher order functions which I think are essentially tied to SLD.

felixyz12• 3 years ago

As I mentioned in another comment, SLG resolution (implemented in XSB Prolog and more recently in SWI-Prolog) does alter the execution model in a way that makes it easier to avoid extralogical constructs. And the wide availability of deeply integrated CLP modules for eg SWI-Prolog also makes it easier to keep programs purely logical.

YeGoblynQueenne22• 3 years ago

> Case in point: basically all your mentioned alternatives (Datalog, SMT, I'd add ASP) have implementations that use the Prolog syntax or something close to it with very similar semantics. But not being tied to that specific execution model can give you (better) answers or more complete answers, termination guarantees, or other things we might need.

To clarify, datalog is a restriction of Prolog excluding functions of more than 0 arguments (i.e. constants). For example, P(x,y) :- Q(f(x), g(y)) is Prolog but not datalog. There's also a restriction in the sharing of variables between head and body literals that ensures termination. The point of datalog is that because of its function-free structure it's decidable. You could simply ground a datalog program and get its entire Herbrand model.

SMT, I don't know much about. It's got nothing to do with Prolog, as far as I know. It's not a programming language, but a framework for boolean constraint solving. I don't know why it (or SAT) is compared to Prolog which has nothing to do with boolean satisfiability problems.

ASP also has nothing much to do with Prolog. It's a strictly first-order language (Prolog is higher-order) that is based on the idea of stable model semantics. Very briefly, a stable model is often the set of queries to which Prolog will answer "true" (or "yes" depending on the interpreter). ASP programs share some syntax with Prolog, but are not definite programs. For one thing, ASP allows classical negation, alongside Prolog's Negation-As-Failure. ASP is said to be well-suited to optimisation problems but to me it's more interesting as a formal treatment of reasoning under uncertainty in a strictly logic framework.

Note that datalog is incomplete (obviously, since it doesn't have functions) and ASP solvers are NP-complete. They are also special-purpose languages, so for example nobody is going to write a web server in ASP anytime soon. By comparison, the SWI-Prolog website runs on a web server written in Prolog (https://eu.swi-prolog.org/dogfood.html).

weinzierl• 3 years ago

That's a very apt description of where Prolog stands. The niche I see it fit in is as a teaching tool for first-order-logic in computer science (not necessarily math). At least that's how I learned it and I'm glad I did.

YeGoblynQueenne22• 3 years ago

Prolog is not a DSL.

Or I guess you can see it as a DSL where the "Domain" is the entire set of programs computable by a Universal Turing Machine. It's a DSL for writing programs that manipulate programs that manipulate programs.

But, not really. It's just not a DSL.

valenterry9• 3 years ago

Just because of performance issues. No other explanation needed really. same reason why it took some time for pure functional programming to take off too. The time for Prolog will come.

discarded10239• 3 years ago

Prolog has many issues; for instance the lack of types and mode declarations, and backtracking almost never being what you want. This was addressed in Mercury [1] but much of the baby was thrown out with the bathwater. (IMHO Mercury ultimately landed closer to the MLs than to Prolog.) The academic literature suggests classic Prolog got fast enough in the late 1980s / early 1990s on stock hardware [2]; see also Peter Van Roy's intriguing book [3] for what happened next.

IMHO it's worth learning about Prolog as a way of looking at the world -- for instance Isabelle and Lambda Prolog are essentially logic programming engines, albeit with more sophisticated terms and unification algorithms.

[1] https://mercurylang.org/

[2] https://www.hpl.hp.com/techreports/Compaq-DEC/PRL-RR-36.pdf

[3] https://www.info.ucl.ac.be/~pvr/book.html

valenterry9• 3 years ago

Right, I have generalized a bit. It doesn't necessarily have to be Prolog, but I'm referring to logic-programming in general.

There is even a new language that allows to embed logic programming inside of "regular" (imperative/functional) programming code. Forgot the name though.

felixyz12• 3 years ago

I agree that the lack of types is a serious deficiency, but surely that can't be the reason Prolog lost steam? I mean... Python?

As for modes, I'm not convinced they are really needed? I do feel they take away a lot of the beauty of Prolog, and a lot of the incentive to think in relations. (Or at least, I fear they do. I haven't actually programmed in Mercury. Would be happy to be proven wrong.)

YeGoblynQueenne22• 3 years ago

Prolog doesn't have typed variables, but any Prolog term has a type, the set of Prolog terms that unify with itself. In practice, when you write Prolog, your arguments to predicates are always implicitly typed, for example if you write a program that takes as argument a list, you'll write something like this:

  a_list_processor([H|T], ..., ):-
    process_list_head(H,Xs)
   ,process_list_tail(T,Ys)
   , ...

Or at least you learn to do that quickly enough. And note that this implicit typing is used both to select clauses in a predicate, and also to index clauses in the Prolog program database by their first term (in which case the first term better be a constant though, not a list). So it's a bit of a bug-repellent, and a bit of an efficiency booster, both. Just like you get with static types.

usrbinbash5• 3 years ago

same reason why it took some time for pure functional programming to take off too.

And when exactly did that happen? And no, I am not about imperative languages incorporating some functional programming stuff. When did a "pure functional" PL become mainstream and widely used in professional software engineering?

Because as far as I can see, I see imperative languages dominating pretty much every field.

valenterry9• 3 years ago

Because as far as I can see, I see imperative languages dominating pretty much every field.

I don't think that contradicts my understanding of "take off".

Alupis2• 3 years ago

same reason why it took some time for pure functional programming to take off too

To be pedantic, almost nobody is using a pure functional language even modernly. Almost all languages that support functional programming are hybrid systems at this point in time, and still have anti-functional constructs such as classes and objects.

I dare say, it would be very difficult to maintain a large system that was purely functional. Some of the concepts revolving around OOP are indeed quite good and lead to better organized, more easily understood codebases.

The future is probably a hybrid approach, not a pure approach as some advocates would prefer.

valenterry9• 3 years ago

Classes and objects don't contract PFP.

Yeah every non academic language (that I know of) has escape hatches.

Still, you can (and many do) write your code in a 99,99% pfp style in e.g. Scala.

usgroup9• 3 years ago

“

“The Japanese Did It” is not the correct answer to “Who Killed Prolog?”. Prolog was killed by the failure in the early 1980s of mainstream AI researchers to find out about Prolog, by their willingness to play along with the “appropriate responses” to FGCS, and by their use of FGCS’s failure as an excuse for continued neglect of a promising alternative to Lisp.

“

Manfred• 3 years ago

It's a natural fit for people with a logic background to validate their ideas computationally. That also meant that people in the 80's and 90's used it for AI research.

For me personally it was a great way to be introduced to different programming paradigms.

It's fine for a programming language to live in a niche.

jillesvangurp• 3 years ago

A lot of the early AI community was using lisp or lisp derivatives. I remember playing with expert systems and Bayesian believe networks in university in the mid nineties. Bayesian believe networks were basically about assigning probabilities to correlations between facts. Early ML became about trying to learn the probabilities.

I also did a Prolog course and I got to teach programming as a teaching assistant in Java to philosophy students that had Prolog as their only programming experience. So, they understood logic but not loops or assignments.

grose2• 3 years ago

I think that Prolog will rise again. If you have an hour to kill, check out this video introduction by Markus Triska about what makes Prolog unique: https://www.youtube.com/watch?v=8XUutFBbUrg

Recently I have been working on getting Trealla Prolog to run in WebAssembly. Support is quite good now, with an easy-to-use JS library[1], Go library[2], and is now one of the best-supported languages for the Spin runtime[3]. I think it's particularly nice for validations, you can write some simple Prolog rules and run the same code on the client and the server now. I would like to explore this more with a dedicated library at some point. Built-in features like DCGs make parsing/generating all kinds of data trivial. I love it.

I think the biggest barrier to Prolog adoption is that it has a really steep learning curve, and most people get burnt by half-assed introductions in college that tick the 'talked about logic programming' box in the syllabus. You really have to think in a completely different way to write a good Prolog program. But, once you get the right mindset, it makes solving complex problems incredibly easy.

I don't really buy the argument that Datalog is enough. Most Datalog implementations I've seen hack on a bunch of extensions to do things like list membership. It would be awesome to see some serious effort put into an efficient Prolog-based database. The closest thing I can think of is TerminusDB, which I've never used but I wish I could convince my company to try it out ;).

My dream is to have some kind of magical serverless persistent Prolog interpreter in the cloud. That's where I hope to take this Wasm stuff at some point. Can you imagine how powerful it would be to use Prolog instead of GraphQL or whatever? Prolog is IMO the perfect language for querying things, with a 1:1 correspondence to how you write the data. It's like SQL without all the ugly bits (I like SQL too, of course). If this sounds cool to you, check out the paper/book on Web Prolog[4] which is a fascinating dive into what a Prolog-based networked query system could look like.

[1] http://github.com/guregu/trealla-js

[2] http://github.com/trealla-prolog/go

[3] https://github.com/guregu/trealla#spin-components

[4] https://github.com/Web-Prolog/swi-web-prolog/raw/master/book...

discarded10239• 3 years ago

There was a lot of work done in the 1980s and 1990s on deductive databases. Somehow it didn't work out. See [1] for a survey from 1995 (coauthored by the most cited guy in CS). There is also the semantic web with its query technology. Also didn't work out.

[1] https://www.sciencedirect.com/science/article/pii/0743106694...

felixyz12• 3 years ago

Amen!

TypeDB[1] is also doing interesting things. Work on incremental view maintenance is also carrying the deductive database vision forward, but under another banner, such as Materialize[2] (based on Datalog, but speaks SQL) or pgsql-ivm[3].

[1]: vaticle.com/typedb [2]: materialize.com [3]: https://github.com/sraoss/pgsql-ivm/

rad_gruchalski4• 3 years ago

You can take a look at OPA Rego. It’s a datalog-inspired language. It will do query and transformation no problem. But it’s not Prolog. Trealla is very nice.

superdisk6• 3 years ago

I think Prolog's main problem is that it presents the idea of "just declare the constraints and it will find the solution," but can't really live up to that. In reality it's an extremely leaky abstraction, and even with pure predicates you need to take its execution model into account to avoid getting stuck in an infinite loop.

I still love the language, but I think there's a point where every learner realizes it's not as beautiful and ideal as they once thought.

usgroup9• 3 years ago

I recently read “Simply Logical” —- it makes very clear that you have to take Prolog’s execution model into account from Chapter 3 onwards.

You really don’t have to go far in Prolog before you have to write a meta interpreter and think about search strategies.

I think this is in plain sight from the beginning.

infogulch2• 3 years ago

Sure, nobody is deceived by the reality of the execution engine. But it's hard not to see a pure statement that is fast in one mode but slow in a different mode, determined entirely by the logically arbitrary order of two clauses, and not wish that the computer could just figure out those details itself.

felixyz12• 3 years ago

The fact that you need to understand the execution model is not a limitation or an accident: it's a feature. One of the aims of prolog is to PROgram in logic. There are plenty of other logical frameworks that you can use to formulate a problem, make deductions etc, but then what do you do? The point of Prolog is that the operational and declarative semantics concur (excuse the non-exact verbiage). People mistakenly think this means there shouldn't be an operational semantics, or that it's a failure if you have to learn about it. If that's what you need, Answer Set Programming (or SMT solvers etc) might be a better fit. But that's not what Prolog is aiming to be.

usgroup9• 3 years ago

I’m not sure it’s a feature that you MUST think about the execution model even for trivial things :-) but that you can and should is certainly a strength.

ghusbands6• 3 years ago

Just try writing (without hacks/cuts) a run-length encoder that produces only one answer ([9,9,9,8,8,7] -> [[9,3], [8,2], [7,1]]) and works forwards and backwards and then tell me the execution model is not a limitation.

It's the sort of thing where people produce versions that seem to work, but fall into infinite loops or inconsistent extra outputs if you go past the first answer. I'm not sure it's actually possible without meta-programming.

yuretz2• 3 years ago

This is the case with any other declarative programming system: while you only care about "what", i.e. the end result matching your declarative specification, you are living a dream. Once you need to fine-tune the details of how you get there (e.g. performance, resource use), or tweak the end result in a way that is not easy to describe within the constraints of your declarative dream land, things get ugly pretty quickly. Query hints in some SQL dialects, vendor-specific hacks in HTML/CSS, Kubernetes YAML templating, etc. are all sad stories about it.

nightowl_games• 3 years ago

Can you elaborate on the flaws of Kubernetes YAML templating? Can you include a description of a better system?

time_to_smile• 3 years ago

I fully agree and came here to post something very similar. Advanced (or even non-trivial) prolog requires such a deep understanding of what prolog is doing under the hood that, at that point, you should be able to implement whatever you're trying to solve just as easily in your favorite non-prolog programming language.

What makes this problem really bad is that everything about your problem that isn't directly related to logic programming is much harder to do in pure prolog. Just reading in a list of facts from a CSV file is non-trivial in prolog.

In the end Prolog makes trivial but hard problems easy, while non-trivial hard problems remain hard to solve, and easy problems, even some trivial ones, also remain hard.

I still love Prolog as well because it really does change the way you think about programming, but it is very far from even Haskell in allowing you to extend these new ways of thinking to solving real world problems.

superdisk6• 3 years ago

Well said. Although there are a few things Prolog excels at, grammars are one of which. DCGs are one of my favorite things about any language ever and I miss them everywhere else.

dboreham• 3 years ago

It was replaced by C++ code in Windows: https://community.osr.com/discussion/264058/re-re-ntdev-re-n...

felixyz12• 3 years ago

A gem!

Animats• 3 years ago

I once wrote something in Turbo Prolong which had both chaining inference and menus. Chaining inference fit well in the language. Drop-down menus were really painful in pure Prolog.

Prolog has a use case, but it's just not that big.

dcow• 3 years ago

Yes.

hyperthesis6• 3 years ago

"Declarative" is sometimes misappropriated by functional programming. Is there a better term? Perhaps explaining it, as in TFA ("specify what, not how")? Or explicitly distinguish from fp? Or some other way?

I think it's fair to say that declarative programming can only work well for "simple" problems (i.e. not requiring turing equivalence), and the strongest wins are when a declarative approach is discerned that didn't seem possible - perhaps by slightly simplifying the problem, to cross this threshold.

discarded10239• 3 years ago

What do you mean by "misappropriated"?

You can try running with what wikipedia says [1] or stack overflow [2]. You could say it's PL marketing terminology from the 1980s and 1990s (FP + Prolog v the world). Roughly these languages aren't based on the idea of changing state, but of course that idea eventually ends up in there somewhere (e.g. monads and laziness in Haskell, DCGs in Prolog).

On the positive side, I'd be comfortable saying that SQL is declarative.

I'd be wary about bringing Turing completeness in here. It seems irrelevant.

[1] https://en.wikipedia.org/wiki/Declarative_programming

[2] https://stackoverflow.com/questions/129628/what-is-declarati...

hyperthesis6• 3 years ago

To answer the first question: "Wrongly taken".

fp is declarative in the shallow sense of being immutable. But you can't just state what you want (e.g. what constraint programming languages enable); you must write a program. If fp is also declarative, what word is left for describing what I mean? Thus: misappropriation.

Not "Turing completeness", "Turing equivalence". Many very simple systems are turing equivalent, so they can exhibit arbitrary complexity. A mapping, from what you have to what you want, is an example of declarative programming. WYSIWYG is declarative. [But arguably this makes "declarative programming" an oxymoron... or we must call html a "programming" language]

Yes, SQL is a lot more declarative; the underlying relational algebra perhaps also is. Something that I'd personally like is a term for systems that enable you to specify the outcome as-is, rather than a composition of steps to achieve it. Perhaps I just want a word for this. I'll give an example:

  (a+b)(x+y) -> ax+bx+ay+by
  (declarative: "fully expanded form")

  (a+b)(x+y) -> a(x+y)+b(x+y) -> ax+ay+b(x+y) -> ax+bx+ay+by
  (steps: "distribute (x+y), distribute a, distribute b")

[1] I agree with wikipedia here: programs describe their desired results without explicitly listing commands or steps that must be performed.

but not in the next sentence: Functional and logical programming languages are characterized by a declarative programming style.

because functional programs specify steps to be performed e.g. g(f(x)) rather than a state of affairs to be attained.

[2] Unfortunately I don't have time to fully read the SO question right now, but I note it's closed and there seems much contention in the comments.

felixyz12• 3 years ago

"Relational programming" pretty much explains what it is, I think. It also makes it clear that functional programming (at least abstractly) is a special case of relational programming, because functions are special cases of relations (namely those relations that are not one-to-many). It's perfectly possible to program in a functional style in a relational language. (In practice, Prolog lacks some of the amenities of say ML-descendant languages, but Curry has them!)

"Declarative" is in my eyes an umbrella term for logic/relational and functional programming, but the word gets used in many ways.

hyperthesis6• 3 years ago

Relational is closer - though I'm personally wanting a term for when you specify the whole outcome in one go, rather than a composition. Like WYSIWYG (see my other reply).

Yes, the word gets used in many ways, that's basically it.

BTW there's injective programming (only one-to-one relations), so programs are reversible.

marcosdumay• 3 years ago

I see the declarative vs. operative concept as a multi-tiered division where none of the general purpose languages can ever be on the declarative extreme, but relational programming is indeed the closest one.

bbwbsb• 3 years ago

I consider declarative and procedural to be readings of a program (i.e, a subjective property). A more interesting property is amenability to formal reasoning, which is what is really wanted.

A distinction from functional programming can be made by noting that in functional programming the input is a proof term, and in logic programming the output is a proof term (evaluation is proof search).

But excluding optimized domain-specific solvers which are making performance/expressiveness/termination tradeoffs, it doesn't seem obvious that programming via proof search has any definite complexity disadvantages in theory (assuming you encode the algorithm in a comparable way), though I admit in practice the existing implementations of logic programming are anemic when it comes to libraries, polish, etc.

xvilka• 3 years ago

There's a nice book[1][2] about Prolog, with modern characteristics. Moreover, there are things like ProbLog[3] and DeepProbLog[4] that allow you to use probabilistic reasoning and power of machine learning. I am personally looking forward for Scryer Prolog[5] to achieve its goals.

[1] https://www.metalevel.at/prolog

[2] https://github.com/triska/the-power-of-prolog

[3] https://github.com/ML-KULeuven/problog

[4] https://github.com/ML-KULeuven/deepproblog

[5] https://github.com/mthom/scryer-prolog

cable2600• 3 years ago

I remember using Borland Turbo Prolog for MS-DOS. It was the bee's knees.

oigursh• 3 years ago

No.

tejtm• 3 years ago

Just a note that for us non-purist-Prolog-curious there is Picat

http://picat-lang.org/

Kuinox2• 3 years ago

I tried to learn prolog. It's hell, I still didn't found a digestible learning material.

The free tooling around it is bad, making you spend too much times debugging simple query as a beginner.

rad_gruchalski4• 3 years ago

I made a few attempts and finally got it by having a go at Google Zanzibar problem using SWI-Prolog. I read TAOP and followed a bunch of videos on youtube. Prolog is nice.

Kuinox2• 3 years ago

I'm not saying prolog is bad, I'm saying I struggled to learn the few things I learned about it, due to the lack of material, or it's bad indexation.

felixyz12• 3 years ago

This is a good take, and historically accurate IMO. Rule engines were indeed marketed as a more practical way of getting the benefits of declarative rules, in a form where supposedly non-programmers would be able to maintain the rule base. And yes, the relational model (in the unattractive guise of SQL) won out in the database space.

Just like functional languages inspire people to write "more functional" code in other languages, I think Prolog can inspire us to think in relations (SQL) and rules. Rules engines grew into monstrosities, but more light-weight rule engines exist and can be a great tool.

zzbn00• 3 years ago

Google's Gerrit code review system had a Prolog engine but I see that it is being depreciated now... https://gerrit-review.googlesource.com/Documentation/prolog-...

Guthur• 3 years ago

Like MLs, like Lisps etc power is trumped by convenience.

Whether it is syntax convenience of C, Java, C#

Or batteries included convenience of Python.

It's a crying shame because we are stuck with the primitive expressiveness of 3rd generation, and are unlikely to break free any time soon if ever. We'd rather get some LLM to generate the reams of boiler plate required to do anything significant.

Jeff_Brown• 3 years ago

hybrid languages that also incorporate some aspects of functional programming or other programming paradigms. Mercury and Mozart/Oz seem to be two of the more prominent ones

Curry also deserves a mention, I think.

https://en.m.wikipedia.org/wiki/Curry_(programming_language)

slavapestov• 3 years ago

The Rust compiler uses a Prolog-like query language internally for type checking generic requirements and traits: https://github.com/rust-lang/chalk

logicmoo4• 3 years ago

If your aim is to develop an SAT solver, it is recommended to write it in Prolog. For more general problem-solving, consider writing an SMT solver or another specialized solver (such as ILP) in Prolog, as it will likely be the most efficient choice.

In the case of creating type-safe software, writing the interpreter in Prolog is a good idea, as it allows for code verification during runtime.

Lastly, if you wish to integrate code and database functionality, it is advisable to use Prolog to develop an enhanced Datalog system, which will offer a more powerful and flexible solution.

mikewarot• 3 years ago

Prolog doesn't have a conventional programmer oriented introduction. You need to at least include a "Hello World" program like this

  main :- write('This is a sample Prolog program'),
      write('\nThis program is written into hello.pl file').
  
  ?- main.
  ?- halt.

Next up, there's the fact that math doesn't use the =, but rather uses is to do evaluations at runtime, which drove me nuts. You have to use foreach or forall to do loops, and there are subtle differences

  ?- forall(between(0,5,N), (A is 5-N,writeln(A))).
  5
  4
  3
  2
  1
  0
  true.

It's just not programmer friendly. At least with spreadsheets, you can see what's going on, Prolog is just weird

It's hard to pick up users, if they can't get enough of a grasp of it to make sense of it.

Why did Prolog lose steam? (2010)

Related Stories