Hacker News Reader

Comment by tannhaeuser | Hacker News Reader

Sure, Erlang was prototyped on Prolog because Prolog has excellent built-in facilities for domain-specific languages: you can define new unary or binary operators along with priorities and associativity rules (you can use this to implement JSON or other expression parsing in like two lines of code, which is kindof shocking for newcomers, but comes very handy for integrating Prolog "microservices" into backend stacks), and you get recursive-decent parsing with backtracking for free as a trivial specialization of Prolog evaluation with a built-in short syntax (definite clause grammars) even.

But apart from syntax, Erlang has quite different goals as a backend language for interruption-free telco equipment compared to Prolog.

View on HN

In The Development of Erlang Joe Armstrong says "We concluded that we would like something like Prolog with added facilities for concurrency and improved error handling".

See pdf linked here - https://news.ycombinator.com/item?id=40998632

You're reading too much into that quote. This is in a section titled "early experiments". It was an initial goal.

There is a lot of historical connection to Prolog, due to the original implementation, and there are syntactic similarities and non-linear pattern matching and dynamic types and a general declarative vibe, but the actual end result of Erlang's evolution, despite the goal of "something like Prolog", is not very much like Prolog at all. Erlang is a functional language, not a logic language. Prolog is a logic language, not a functional language. General goals like in that quote can change over the decade-long development of a language.

You are stating some obvious things which are not what is being argued here. I only gave the above quote because it is the conclusion that the Erlang designers came to after a lot of research/playing/implementing with other languages (read the full paper and others listed below). I had also stated above that "he conceived Erlang as Prolog-Ideas+Functional/Procedural+Concurrency+Fault-Tolerance." So we already know Erlang is a different language. Note that Robert Virding also states in the Erlog page "due to the fact that Erlang evolved from Prolog". So obviously Prolog was a huge influence on Erlang design (not to be confused with the fact that the first experimentation was done in Prolog) in addition to other systems/languages. The above paper also states "It was a strange mixture, with declarative features (inherited from Prolog), multi-tasking and concurrency (inherited from EriPascal and Ada) and an original combination of error handling mechanisms". The last came from AXE/PLEX and others.

Joe Armstrong wrote two papers The Development of Erlang (linked here https://news.ycombinator.com/item?id=40998632) and a longer A History of Erlang (pdf at https://www.labouseur.com/courses/erlang/history-of-erlang-a...). In addition to his thesis (pdf at https://erlang.org/download/armstrong_thesis_2003.pdf) they provide a fascinating study into what goes into the design of a language i.e. lots of messy experiments, shifting goals, inspiration/features from many different languages etc. until everything coalesces into a organic whole which is then validated by users. Reading the above two papers will give you a more complete picture of Prolog's influence on Erlang (in addition to others).

Thanks, I know that history, and nobody is denying the influence. The broader context are your claims about "a lot of commonalities here". There are commonalities, but they are shallow.

That we have to disagree. Leaving aside the non-Prolog features, much of the Erlang syntax is definitely inspired by Prolog (mentioned in many papers including https://www.erlang.org/faq/academic). That was the reason the HN submission i had referenced which linked to a reference sheet comparing Prolog/Erlang/Elixir was so interesting (direct link https://hyperpolyglot.org/logic). In most cases the Erlang syntax is just a simplified version.

I mentioned the syntactic similarities (and more) above. Syntax is shallow.

Yeah but to be honest, Erlang ended up being not something like Prolog at all.

I think Joe Armstrong was a user here and I interacted with him waaaay way back when I first joined. He's dead now :(

See https://news.ycombinator.com/item?id=41013798

I saw, as I saw your upthread quote from the paper above but I think you misunderstand it. The previous paragraph gives context to your quote:

  The main conclusion [5] was that declarative language
  programs for POTS were a lot shorter and easier to
  understand than imperative language programs. Un-
  fortunately the declarative languages lacked features
  for concurrency control and had poor error handling
  facilities.

  We concluded that we would like something like Pro-
  log with added facilities for concurrency and improved
  error handling. No such language existed at the time.

"Something like Prolog" here reefers to the declarative features found in Prolog as well as other declarative languages, left unnamed in the quote, which make programs "a lot shorter and easier to understand thatn imperative language programs".

But that's where the "commonalities" you mention in your OP, between Prolog and Erlang, end. Similarly, CSS and XML are declarative but that's where their "commonalities" with Prolog (and Erlang) end.

I'm insisting with this because I'm concerned that your comment promulgates a common misconception about the, like you say, "commonalities" between Prolog and Erlang. These end with syntactic similarities and misunderstanding this can cause some disappointment to people trying to go from one to the other. I've seen a similar misunderstanding arise about "commonalities" between Prolog and, e.g. Haskell - they both have weird, arcane syntax and immutable data structures, but that's all. Or, think of C and javascript: both Algol like syntax and in fact the first js compilres must have been written in C, but that's all. etc.

There is much, much more to Prolog than the declarative syntax.

First, my other comments for reference; https://news.ycombinator.com/item?id=41013798 and https://news.ycombinator.com/item?id=41015406

I had explicitly stated above that "he conceived Erlang as Prolog-Ideas+Functional/Procedural+Concurrency+Fault-Tolerance." Here i am not making an equivalence between Prolog and Erlang but emphasizing the inspiration that Prolog provided for the design of Erlang (that is what is to be understood from all the links to the papers and comparison charts given in earlier comments). That inspiration is in the syntax of pattern matching, Atoms/Variables, Module/Export directives etc. Similarity of Syntax is very important for learning/understanding new languages since your cognitive load is decreased dramatically. A good example is C++ to Java/C# where the similarity in syntax (though the runtime object model is very different) is what was crucial to the widespread adoption of the latter. It is in this sense that Prolog to Erlang comparisons should be understood.

Your arguments of grouping/comparing widely dissimilar languages are somewhat disingenuous. A much better side-by-side comparison of languages is https://hyperpolyglot.org/ where the author has tried to group by intended functionality and historical development. I think this is a good way to do it.

>> It is in this sense that Prolog to Erlang comparisons should be understood.

What I'm saying is that this is the entirely wrong thing to focus on, i.e. syntactic similarities. Prolog is a language of the logic programming paradigm, one of the first ones. Its syntax is that of (a restriction of) the first order predicate calculus. Its programs are First Order Logic theorems. Its interpreter is an automated theorem prover. The motivation for Prolog is the ability to program a computer using the syntax and semantics of First Order Logic, to be able to prove the truth values of statements in a formal language automatically, with a computer. It has nothing to do with Erlang, a functional programming language designed to program telephony switches. Any similarity is superficial: Erlang expressions are not definite clauses; everything in a Prolog program is a definite clause, modulo punctuation. Knowing Erlang will not help you learn Prolog just because the syntax looks similar. You can forget about that right now- that's the misconception I'm trying to correct. Don't encourage people to try to understand Prolog that way because you will only cause them pain.

Take the site you link to as an example. It tries to bodge together Prolog with Erlang and Elixir in a "side-by-side reference sheet" that includes rows for "assignment" and "parallel assignment" using =/2. That's the unification predicate! Prolog does not have assignment! Just imagine the suffering of a novice programmer trying to use their knowledge of assignment, in Erlang or any language, to understand the following:

  ?- X = a, X = b.
  false.

That's just setting up programming students to fail, to fail to understand, to fail to learn- and to only succeed in blaming Prolog for being a stupid, painful language that is hard to learn. Of course it's hard to learn! If you go around telling people that they can learn it more easily if they know Erlang!

>> Your arguments of grouping/comparing widely dissimilar languages are somewhat disingenuous.

No, the point is that they are widely dissimilar and that you won't understand that if you just stop at the syntax. Like the site you link to, where someone clearly made an effort to memorise syntax but completely failed to understand semantics.

Semantics shmantics! The attitude I see here is the one that Peter Norvig criticises in "Teach yourself programming in 10 years": try to find a shortcut around the hard stuff so you don't have to use the brain. You don't learn anything that way.

You are on a completely different tangent here. To repeat myself, we are not making a equivalence comparison between Prolog and Erlang. That their computation models are different is well-known. The OP is already studying Prolog and therefore has some familiarity with its computation model. He/She is asking for advice on what more to study. What is being suggested is that since Erlang was inspired by Prolog (to whatever extent) it would be useful for the OP to study that too since it has more usage/acceptance in the industry given the concurrent/distributed architectures in vogue today.

> Knowing Erlang will not help you learn Prolog just because the syntax looks similar.

That is not what is being suggested. This is your fundamental misconception which permeates the rest of your post. The OP is already studying Prolog and is being advised to look into Erlang too in addition to that. It is not "instead of" but "in addition to" (due to the history of Erlang). Furthermore Erlog allows you to embed Prolog within Erlang giving the OP the best of both worlds.

> you won't understand that if you just stop at the syntax.

Again this is your misconception. That is not what is being suggested. Nobody in their right mind will say "stop at the syntax". But syntax is the gateway into the study of computer languages i.e. you first show the syntax and then explain its semantics. Humans learn new things using analogies/similes/metaphors/etc. Here similar syntax is a great help since it eases the cognitive load while learning new concepts. When the semantic model varies, the dissonance may not be too great and so we can better modify our mental models and understanding.

The website listing the syntax comparisons of various languages is actually pretty useful when looked at with the above viewpoint. Start with your known syntax in a language you know, see how it maps to the same/similar syntax in the language you want to study and then lookup/compare/contrast the semantics of it in both the languages. It is like a cheat-sheet while studying a proper programming language textbook.

>> You are on a completely different tangent here.

I don't think so. In your comment above you keep saying that it's not about syntax, then immediately after you switch to explaining how it really is. Notice this for instance. I say:

>>> Knowing Erlang will not help you learn Prolog just because the syntax looks similar.

You say:

>> That is not what is being suggested.

Then you go on to say:

>> But syntax is the gateway into the study of computer languages i.e. you first show the syntax and then explain its semantics. Humans learn new things using analogies/similes/metaphors/etc. Here similar syntax is a great help since it eases the cognitive load while learning new concepts.

And:

>> Start with your known syntax in a language you know, see how it maps to the same/similar syntax in the language you want to study and then lookup/compare/contrast the semantics of it in both the languages. It is like a cheat-sheet while studying a proper programming language textbook.

So it's hard for me to see how you're not suggesting that syntax will help you learn Prolog, when you're arguing that syntax helps you learn a new language.

And what I'm saying is that knowing something about the syntax of Erlang will not help you learn anything about Prolog. I don't think you can accept that at all, but it's true and it's my experience of many years watching people, including myself, trying to learn Prolog using their knowledge of other languages, either their syntax or semantics. My experience includes working as a teaching assistant for a Prolog course during my PhD. The truth is that Prolog is not an easy language to learn because it works very differently than any other programming language and you won't be able to use your knowledge of any other programming language to learn it. If you want to learn Prolog you have to start by finding a way to set aside most of what you know about programming. That is very hard to do, and so it's very hard to learn Prolog, except at a very superficial level.

I appreciate that's not easy to accept without long time experience.

But I don't think we're getting anywhere with this conversation so I'll respectfully bow out and thank you for the patient and civil exchange of views.

Again, the OP is moving from Prolog -> Erlang and not the other way around. You keep repeating "So it's hard for me to see how you're not suggesting that syntax will help you learn Prolog," but that is not OPs path. Erlang evolved out of Prolog and hence some of the syntax/semantics are very similar.

To settle this once and for all; in your previous post you gave the example of assignment which actually proves my point that you can easily go from Prolog to Erlang. Why? Because the behaviour is the same in Erlang i.e. there is no assignment but only pattern-matching. You can open a Erlang shell and type in X=a, X=b and it will barf. To show it even more convincingly type in Y=1+2, Y=3 and Y=4 and it will barf only on the last line. Lhs=Rhs are pattern-matched and variables are single assignment. A person with prior exposure to Prolog already knows this and hence can easily map it to Erlang. Only people coming from imperative languages (eg. C) need to adjust their mental model of the "=" operator. When i started with Erlang years ago that was what i had to learn first but thankfully all Erlang books cover it in the introductory chapter itself. Also functions in Erlang are written as a series of clauses which are pattern-matched.

I presume you do not have experience with Erlang and so your experience with only Prolog blinds you to the similarities. I highly recommend you get a Erlang book (Joe Armstrong's book is a good one since he uses much of the terminology like "terms", "clauses" etc. from Prolog) and with the help of the cheatsheet just try out the basic syntax in the Erlang shell and see what maps to Prolog. I am quite sure that you will find it convincing.

Finally w.r.t. Prolog's computation model being based on Predicate Calculus/First-Order Logic and so the model/semantics are very important is not being denied. I myself came to Predicate Calculus from the "Program Correctness" viewpoint of Floyd/Hoare/Dijkstra and hence an quite aware of its intricacies. But for most programmers without such a background in Set theory/Relations/Logic model building takes time and so syntax is their entry point and understanding happens over a period of time with study and trial-and-error.

Reply on news.ycombinator.com