Hacker News new | past | comments | ask | show | jobs | submit login

TLA+ is only "executable" in the same sense that an algebraic expression is executable. It's perfectly possible to write things on TLA+ that can not be simply executed linearly. (These overlap to a great extent with the things which TLC rejects.) As a basic example, it's easy to write a statement with \A (unbounded universal quantification) whose truth can only be judged by a proof engine.

Specification languages are explicitly not programming languages, for the core reason that programming languages dictate only what must occur; whereas specification languages can dictate what must not occur. It's not possible with a "specification" written using a programming language to determine what of a program is actually the specification, vs. what is an accident of the implementation.




Being able to create systems by writing specifications and having the computer figure out how to execute them was basically the point of fifth generation programming languages.

More relevant today, you can execute other "specification" languages like Coq and Idris because they support things outside the narrow feature set of specification usecases.

TLA+ isn't executable and doesn't look like an imperative language because the authors don't want it to be, not because there's some universal line dividing specification languages from programming languages. It's also one of the biggest hurdles to TLA+ usage.


> Being able to create systems by writing specifications and having the computer figure out how to execute them was basically the point of fifth generation programming languages.

Yeah, but in maths you can specify anything, including things that the computer is unlikely to figure out how to execute if it's possible at all. Programming languages of every generation are very useful, as is mathematics, even though they're not the same thing.

> More relevant today, you can execute other "specification" languages like Coq and Idris because they support things outside the narrow feature set of specification usecases.

Coq and Idris are very different in the way they're typically used, and I'd say TLA+ is much closer to Coq than to Idris (and is probably more popular than the two combined), but to "execute" anything the specification needs to be at a certain level that's detailed enough to produce a program, and oftentimes that is very much not what you want.

It would be extremely useful to have a language that you could describe the various properties of a car and it would compile your specification into the design of a car (you would need to give it sufficient detail as there are choices to be made). But it would also be extremely useful to have a language that could be used to learn certain things about a car -- say, it's braking distance -- without specifying it in sufficient detail to actually build one. That is what maths is good for -- describing things at arbitrary levels of detail to answer relevant questions.

For example, you may have a 5 MLOC distributed system, and you want to know if a certain kind of failure may lead to data loss. You could use TLA+ to describe just the relevant details to answer the question in, say, 200 lines of formulas. That you cannot compile those formulas into a working 5 MLOC piece of software is not a downside of mathematics, but rather the point.

> TLA+ isn't executable and doesn't look like an imperative language because the authors don't want it to be

And because it's not a language for programming but a language for mathematics, so it looks and feels pretty close to plain mathematics (only it's formal), as that's the obvious choice for writing mathematics.


> in maths you can specify anything, including things that the computer is unlikely to figure out how to execute if it's possible at all.

Well, in TLA+ you can write programs that run forever (or at longer than you'll live) and don't do anything like "model check" or whatever you want to call executing TLA+, even though they are perfectly sound mathematically. This should make it clear that TLA+ is not maths.


I don't understand the order of your implication. You can use maths (say, some ZFC formalism) to specify the execution of any Python or Java or C program and even write a software tool that executes it. It's the converse that isn't true: You cannot write a Python or Java or C program that accurately expresses many mathematical theorems (e.g. you can only express computable numbers in a program). I.e. the expressivity of mathematics includes that of programming, but not vice versa.

In TLA+ you can express every theorem in ZFC, but that doesn't mean you can automatically prove or disprove every proposition or even every theorem, because that is indeed a limitation of mathematics. There are also lots and lots of theorems you can state and prove in TLA+ yet not prove automatically with the TLC model-checker (or, indeed, with any known automatic proof method). That is a limitation of TLC (or of any known automatic proof methods), but not one of TLA+.


Actually the converse is true: Lean and other projects have formalized most mathematical theorems. But it is an "additive" process - it is easy to paraphrase a formal Lean theorem as colloquial mathematics, but it is hard to formalize colloquial mathematics in Lean. Some of this is due to Lean not being as developed as it could be, but also there is simply that some "theorems" in mathematics are simply "wrong" in that they make unstated assumptions and handwave away important parts of the proof. It is in this sense that programming is less expressive than mathematics, in that you can get away with writing things in mathematics that you can't get through a theorem checker. And this is conversely why I say that programs for theorem checkers are executable - the requirement to pass a theorem checker imposes constraints on proof structure and such that is not found in the "natural" language of mathematics. The lack of these limitations is what I would say the limitation of mathematics is, that even well-known proofs are not necessarily completely "true" due to unstated assumptions.

Now regarding TLA+ vs TLC, I am not clear what the utility of a TLA+ program that cannot be checked with TLC / TLAPS / etc. When you say "prove" a TLA+ program I first thought this was formally checking it with TLC / TLAPS / etc. But it seems you have a different notion of proof, some sort of handwaving "it looks right" notion. From my perspective this reduces a TLA+ program to a piece of writing, since nothing automated can be done with it. You might as well say "You can express every theorem in ZFC in Java by writing it in a comment" - it is not an informative observation. The interesting TLA+ programs are the ones that can be checked with TLC / TLAPS / etc., and to the extent one can work with these programs programmatically, TLA+ is a programming language.


> And this is conversely why I say that programs for theorem checkers are executable - the requirement to pass a theorem checker imposes constraints on proof structure and such that is not found in the "natural" language of mathematics.

But TLA+ formulas not programs for a proof checker. There is a model checker than can check some subset of TLA+ and a proof checker that can check some TLA+ proofs (if they're detailed enough to be checked), but you can even write proofs in TLA+ that TLAPS cannot check.

> Now regarding TLA+ vs TLC, I am not clear what the utility of a TLA+ program that cannot be checked with TLC / TLAPS / etc.

First, there is no such thing as a TLA+ "program". You write mathematical definition and state propositions. Second, that's like saying that it's not clear what the utility of any formal mathematics without tools, but formalisms were invented long before there were computers.

You can prove things without having them checked by a proof checker -- indeed that is how nearly all mathematical theorems have been proven -- and it's a rigorous process that isn't at all the same as things just "seeming" right.

Obviously, most of the practical usage of TLA+ involves use of TLC, some smaller amount involves the use of TLAPS, but whether or not you think there is utility in mathematics without mechanised tools, it doesn't change the fact that TLA+ is formalised mathematics.

> From my perspective this reduces a TLA+ program to a piece of writing, since nothing automated can be done with it.

It's not a program, and often times mathematics is "a piece of writing that nothing automated can be done with."

> The interesting TLA+ programs

Not programs, formulas. Again, they cannot be executed, although their veracity could be checked. Here is a TLA+ specification: `3 * 3 = 9`. Here is another one: `∀ x ∈ Nat : x * x ≥ x` [1]. These two can easily be checked to be true, yet they are not programs. Sure, mathematics can (and is) used to describe aspects of the dynamics of a thrown ball or of an algorithm, but that doesn't make mathematics a ball-manufacturing machine nor a programming language. You can use TLA+ to specify the quicksort algorithm and prove that it, indeed, sorts its input, but you don't use it to produce an executable program that sorts a list (unless you use a very limited subset and write a tool that can translate formulas of certain forms to a computer program).

[1]: I'm eliding some details because TLA+ formulas are interpreted in the TLA logic, where formulas have very specific semantics, but it's close enough for our purpose.


> It's not a program

It is a program when I can do java tlatk.TLC someprogram.tla.

You say `3 * 3 = 9` is a TLA+ specification. Well, here is a Prolog program: 3*3 #= 9. Is there a difference? No. The output when I run the Prolog program? "yes". The output when I run the TLC program? I haven't tried, but it is probably similar to "yes" or "no". It is in this sense that you can run TLA+ programs and get a relatively small output of whether it checks. Maybe you don't consider this programming, but people have done more with less, e.g. lambda tarpits where all that happens is lambdas reduce to more lambdas. In contrast the value space of TLA+ is quite rich, it is only the usability of it that is limited because Leslie Lamport continues to insist that TLA+ is "not a programming language".


> It is a program when I can do java tlatk.TLC someprogram.tla.

TLC is a program that takes TLA+ formulas as input and produces an output. I have a program that reads newspaper articles out loud. Doesn't make those articles programs or their authors programmers.

> Is there a difference? No

You're using implication in the wrong direction. Some mathematics can indeed appear in a programming language, but the question is not whether there's something in TLA+ that could be in Python or whether you can specify a Java program in TLA+ -- of course you can. The question is whether there's something in TLA+ that cannot be in a programming language, and, indeed, there is. All Prolog programs are programs; you can run them. Not only can you not many TLA+ specifications (or mathematical formulas in general) in a similar way to a program, many mathematical formulas cannot possibly describe any program that's implementable in the real world. You can write mathematical propositions about non-computable real numbers; you can specify a decision oracle for the halting problem (and because you can do it in mathematics, you can write it formally in TLA+).

Similarly, the fact you can specify orbital dynamics in maths doesn't make it physics, and it's easy to see it's not physics because you can just as easily specify motion that breaks physical laws.

> Maybe you don't consider this programming, but people have done more with less

You can certainly run computer programs that tell you interesting stuff about a TLA+ specification, and you can even say that the existence of such programs is the source of much of the value in TLA+. But you can run interesting and useful programs that do stuff with images, yet however you want to look at it philosophically, most photographers and programmers would agree that photography and programming are two pretty different disciplines even though photographers may use software in their work.


You can write hello world in TLA+:

Main == PrintT("hello world")

You can't write much else, because print is the only implemented command (no input commands even), but it's clearly not just a program that reads newspapers out loud. There is an execution semantics and so on, as found in typical programming languages. It would not be hard to add input, implemented as a "hack" along the lines of print, although of course TLA+ is nondeterministic like most logic programming languages so there is some tricky semantics.

I'm not saying TLA+ is a good programming language - it is more along the lines of Brainfuck or TeX, "use it only out of necessity or masochism". But at least in my book it is undeniably in the category of programming languages.


> You can write hello world in TLA+:

That's not a good example, but before I get to that, you are again using logical implications in the wrong direction. You can describe physical systems, including programs in mathematics. What makes mathematics not the same as physics or programming is that you can also describe things that aren't physical or computable. It's like you're trying to prove that New York is the United States by pointing out that if you're in NY you're in the US.

Let me write that another way: programming ⊊ mathematics.

So you can write all programs in mathematics; you cannot do all mathematics in a programming language. Therefore, to know whether TLA+ is programming or mathematics the question is not whether you can describe programs in TLA+ but whether you can describe the maths that isn't programs, and you can.

Now, the reason that your example is not good is that the PrintT operator is defined in TLA+ [1] like so:

    PrintT(x) ≜ TRUE
It has no meaning in TLA+ other than the value TRUE. It's just another way of writing TRUE. The computer program TLC, however, prints some output when it encounters the PrintT operator in a specification, but that behaviour is very much not the meaning that operator has in TLA+. TLC can equally print "hello" whenever it encounters the number 3 in a mathematical formula. There are other TLC tricks that can be used to manipulate it to do clever stuff [2], but that's all specific to TLC and not part of the TLA+ language.

You could, however, have pointed out that you can specify a Hello, World program in TLA+, only that is unsurprising because you'd specify it in mathematics, and mathematics can also specify a Hello, World program. Here's how I would specify such a program in TLA+ (ignoring liveness for the sake of simplifying things):

    VARIABLE console
    Spec ≜ console = "" ∧ □[console' = "Hello, World!"]_console
It means that a variable named console (which we'll take to represent the content of the console) starts out blank, and at some future point its content becomes "Hello, World!" and it is not changed further.

You could also specify it another way, say, represent the console as a function of time (I'll make it discrete for simplicity):

    console[t ∈ Nat] ≜ IF t = 0 THEN "" ELSE "Hello, World!"
[1]: https://github.com/tlaplus/tlaplus/blob/0dbe98d51d6f05c35630...

[2]: I've even given a talk about such tricks: https://youtu.be/TP3SY0EUV2A


> the PrintT operator is defined in TLA+ like so

No, it is defined like so: https://github.com/tlaplus/tlaplus/blob/0dbe98d51d6f05c35630... This is the same strategy that Haskell uses for its primops, a placeholder file and implementation in another language. I guess you didn't understand my point that it would be easy to extend TLA+/TLC to more primops, like memory, a Java FFI, and so on, making it a fully featured programming language. I don't care what "the TLA+ language" is, C++ implementers regularly toss the spec in the dumpster and just like you have C++/LLVM and C++/GCC there similarly is a dialect of TLA+ for each implementation.


> No, it is defined like so

No, you are looking at the TLC source code, not TLA+ definitions. If you read the TLC documentation, it makes it very clear that it is not TLA+, but a software tool for model checking a certain subset of TLA+ and has some programming-like features that are not in TLA+. Other TLA+ tools also focus on other TLA+ subsets and have their own features, and they're not necessarily consistent with one another.

> This is the same strategy that Haskell uses for its primops

It really isn't, because TLC isn't the TLA+ compiler/interpreter. In TLA+ PrintT is just synonym for TRUE. We can talk about TLC if you like, but you shouldn't confuse the two.

> I guess you didn't understand my point that it would be easy to extend TLA+/TLC to more primops

TLA+ and TLC are two different things. TLA+ is a formal language for writing mathematical specifications. TLC is one of the several tools for analysing TLA+ specifications written in a subset of TLA+.

> I don't care what "the TLA+ language" is

That would sort-of made sense if TLC was the canonical tool for analysing TLA+ but it isn't (certainly not in the sense that ghc is the canonical Haskell compiler). In SANY, TLAPS, and Apalache (another TLA+ model checker), PrintT is just TRUE. In our work on the TLA+ toolbox, we're always careful to separate what's TLA+ and what are the features of the software tools, and in the materials we produce about TLA+ and the TLA+ tools these differences are explained.

In other words, you -- as someone who doesn't know TLA+ -- may not care about what TLA+ is and what TLC is, but those who use TLA+ very much care (though maybe not in the first week), because that distinction is required to put the language to good use.

If you choose to learn TLA+, I'm confident you'll come to agree. Everything I've said here will become clear once learn TLA+, TLC, and TLAPS (or even just TLA+). Those who actually use TLA+ very much do care what "the TLA+ language" is and what are some TLC features, because they often want to use multiple tools to do different things with their specifications, and PrintT only prints stuff when you're using TLC, not, say, when you're using TLAPS or Apalache. Once you gain experience with TLA+ you understand which subset of it can be checked by TLC and which isn't, and you write certain specifications with the intent of checking them in TLC and others for other uses, as you know that they fall outside TLC's subset of TLA+.

I think this short talk I gave and I linked to before (https://www.youtube.com/watch?v=TP3SY0EUV2A) gives a sense of the different ways you use TLA+ and how you combine them. It shows how you use the language as you do ordinary maths -- by manipulating formulas on paper -- and then use the results in conjunction with TLC to obtain some particular kind of automation. In fact, it is about how to use TLA+ to rewrite some TLA+ specifications so that they could be checked in some interesting way by TLC, i.e. the whole premise is that TLA+ users are confronted by the differences between TLC and TLA+, and it shows how to use certain deductions in practice to put TLC into some interesting uses for kinds of TLA+ specifications that TLA+ users would normally think fall outside the bounds of TLC. Anyway, anyone who actually uses TLA+ is -- and needs to be -- very much aware of the differences between TLA+ and TLC, recognises their different roles.


> If you read the TLC documentation, it makes it very clear that it is not TLA+.

Fine, clearly you are missing the point I am making about how languages become confused with implementations. Just s/TLA+/TLC/ in all the above. Is TLC a programming language implementation or not? Consider for example https://github.com/will62794/tlaplus_repl which evaluates TLC expressions. At what point is there sufficient programming language functionality for you to become convinced that TLC is a programming language?


TL;DR: Of course TLA+ can be used to describe all programs, as can all sufficiently-rich mathematical formalisms (regardless of TLC; TLC has got nothing to do with that). It is definitely not a programming language because its expressive power comes from its ability to describe things that cannot be computed (or practically computed). I.e. it's not a programming language not because it's too small of a subset of programming, but rather a vastly larger superset of programming. In other words, TLA+ (and mathematics in general) is not a programming language not because it is less than a programming language but because it is more. If TLA+ is a programming language, then so is any rich mathematical formalism, any drawing tool, or English for that matter.

> Is TLC a programming language implementation or not?

Ok, so first of all, TLC is not "an implementation of TLA+" and not because it can only check a limited subset of TLA+.

To sho that, let me take a short but important detour. What is the purpose of a computer program? It is to produce an output, either for yourself or for others you give the program to. In contrast, what is the purpose of a TLA+ specification, i.e. what is it that you do with it, or what is the deliverable? Clearly, it's not some output because a TLA+ specification has no output (I'll get back to TLC in a moment). Once you're convinced that the TLA+ specification fulfils the propositions you're interested in -- either by inspection, formal manipulation and proof on paper, formal proof checked by TLAPS, or a successful run of TLC -- the deliverable is the TLA+ specification itself, the set of formulas, whose purpose is then for someone (maybe yourself or someone else) to use as a design for some system to build -- a program, a computer chip etc.. So a the purpose and deliverable of a TLA+ specification is it itself, just like the purpose of an architectural blueprint.

Now, back to TLC. TLC is, no doubt, a program that takes as an input a TLA+ specification written in a particular subset of TLA+ and some additional configuration that defines the boundaries of state space to model-check, and produces an output that's either TRUE or a counterexample. There are other modes of running TLC, as well, for which special operators like PrintT can be useful.

Note that there are many other programs that do something with a piece of mathematics and produce an answer. The simplest one is a calculator. A calculator is a program that takes as an input some portion of a mathematical statement in a small subset of mathematics (arithmetic on some finite subset of integers and rationals) and produces an output.

So we can take a partial mathematical statement 3 + 4 and use it, in conjunction with a calculator, as a program that produces the output 7.

So finally, we can answer your question. TLC is a program that's more sophisticated than ordinary calculators, and we can certainly write some mathematical statements in a subset of TLA+ so that when we feed them into TLC we create some interesting output. It is similar to a CAD/CAM tool you run on an architectural blueprint. But is the tool "an implementation" of the blueprint? I don't think that makes sense.

> At what point is there sufficient programming language functionality for you to become convinced that TLC is a programming language?

TLC is not a language but a tool that can process a subset of TLA+. The existence of calculators or of TLC does not mean that mathematics is a programming language, because:

1. Even though mathematics can be used to describe all programs and all of physics -- because the domain describable by mathematics, and therefore by TLA+ -- is a superset of all the behaviours of physical and computable things, it is also a strict superset, and mathematics (and therefore TLA+) can describe lots and lots of things that cannot possibly be realised by anything in the physical world or by any computation.

2. Even though some subset of mathematics (and therefore of TLA+) can be used in conjunction with some program to produce an intended output, that output is not the purpose of mathematics (and therefore of TLA+).

As for 1, you may then ask what's the point of a language that is ultimately intended to produce designs for physically-realisable systems to encompass all of mathematics and describe things that are not realisable. There are two answers to that: first, it makes the working with the language much simpler, just as working in classical mathematics is simpler than working in constructive mathematics (which is based on intuitionistic or some other constructive logic rather than classical logic). Second (and this is really an application of the first answer), specifying non-realisable things is helpful when specifying properties of realisable things. For example, suppose you design an algorithm that can decide whether some specific subset of programs halt. You want to check the proposition that:

    InMySubset(InputProgram) ⇒ MyAlgorithm(InputProgram) = Halts(InputProgram)
and to do that you need to describe the operator Halts even though it is not realisable (as it's not computable). So Halts clearly cannot be written as a program, yet defining it in some mathematical formalism is needed to express a real property of a real program. (BTW, the definition of a Halts operator is a trivial one-liner in TLA+ [1]).

I see that again and again you're getting stuck on the point that because there are programs that can evaluate mathematical statements, mathematics is a programming language. But the fact that you can describe any program or any physical system in mathematics is the point and power of mathematics. But mathematics is neither physics nor programming because it can also describe things outside the world of programs and physics.

Of course you can write programs in every rich mathematical formalism, including TLA+ (and TLC has nothing to do with it; this would be true even if TLC didn't exist). But programming languages are languages that can describe things that live in a small subset of the world of mathematics. TLA+ is not a programming language not because it doesn't contain the universe of all programs -- it does; every imaginable program could be specified in mathematics and specifically in TLA+ -- but because it contains much, much more.

So if you want to define "a programming language" as any language in which you could describe many or perhaps even all computations, then every rich mathematical formalism (and therefore TLA+) could be considered a programming language. But I would think that a language where most things you write are not computable isn't a programming language. Rather, a programming language is one where everything you write in the language is at the very least computable, and that is certainly not the case for TLA+.

Something similar is true for English. You can use English to describe any conceivable algorithm, and there are now tools that could convert a subset of such descriptions to executable software. But the fact you can use English to describe programs doesn't make English a programming language, because many of the things it is used to described aren't programs.

But even if you insist that mathematics (or English) is a programming language, the point is still that mathematics (and TLA+) exists to describe useful things that go well beyond what could be described in a programming language [1].

[1]: Halts(Program, vars) ≜ Program ⇒ ⬦□(vars' = vars)

[2]: I'm overlooking the type level in programming languages with rich type systems (like Agda), but the point still stands, only the details are more technical.


This thread has been going on, let me try to distill the points:

- TLA+ is "definitely not" a programming language (per you and Leslie Lamport).

- TLC has got nothing to do with TLA+ (as a mathematical formalism). TLC is not "an implementation of TLA+". (per you)

- TLC is a tool that can process "something like" TLA+. You say "subset", but it seems to me it is not a strict subset, because special operators like "Print" have different semantics. Let's suggestively call what it processes "TLA-PL". You mention additional configuration but the configuration can be empty so it's really like a pragma or compiler option.

- TLC can evaluate and print TLA-PL expressions in a REPL. (per the repo I linked)

- TLC and TLA-PL could be extended to implement typical programming language primitives such as input, a Java FFI, etc., fairly easily (per observation of the source code)

- TLA-PL is not TLA+, because it is not a rich mathematical formalism, like a drawing tool or English. The purpose of a TLA-PL document ("program") is to produce an output that's either TRUE or a counterexample, although there are other modes of running TLA-PL. In contrast, the purpose of TLA+ is itself, and a TLA+ document ("specification") has no output - the deliverable is the document.

Now it is true that other programs have REPL-like functionality, like the calculator you mention. Generally the benchmark between calculation and programming is Turing completeness, e.g. whether the language can express recursion. In a calculator, if you add a few statements like stack push/pop and command names, suddenly it is a "programmable" calculator like the HP-32S, and Turing complete, and the calculation language becomes a programming language. What about TLA-PL? Naturally TLA-PL expresses recursive statements easily - it is almost trivially Turing complete and hence a programming language. And it is clear by definition that TLC is an interpreter for TLA-PL, so TLA-PL is even an implemented programming language. This is what distinguishes it from the majority of formalisms, in that most formalisms (English, mathematics), although potentially usable for programming, do not have working implementations. It is not a requirement to be a programming language that everything written in the language is computable - Verilog, for example, is actually quite flexible as a hardware synthesis language, allowing one to write unsynthesizable programs, but in practice people simply avoid writing these programs when doing hardware synthesis. Similarly I am sure that valid-looking TLA-PL programs will look correct but nonetheless fail to run under TLC due to limitations of the model checking and so on.

Now it is true that TLC, although it implements TLA-PL, is not an implementation of TLA+, as by definition TLA+ is like mathematics, infinite in scope, hence not implementable. I would argue this also means TLA+ also isn't even definable, but that's a separate issue. Similarly, Leslie Lamport's purpose in creating TLA+ was not (and is not) to create the programming language TLA-PL, even though it exists. This to me is what you're getting stuck on. As a programming language designer, what I care about is TLA-PL. To me it is clear as day that TLA-PL exists as a programming language and and could be turned into a useful one given sufficient effort to modify TLC. In contrast, all I hear from you is "TLA+ this", "TLA+ that", "pay no attention to the working implementation of TLA-PL". But as I said, I don't care about TLA+ - as soon as you say it realizes unrealizable things, you are speaking poetry rather than programming language design. There are tricks like lazy evaluation and so on where a computer represents "unrepresentable" objects symbolically and thus can manipulate them, and from my understanding some of these tricks are implemented in TLC and TLAPS, but it seems clear you are talking about a level beyond this, where a TLA+ specification cannot be evaluated even with symbolic tricks.


I think your question really is, could one design a programming language -- i.e. a language where everything is executable -- inspired by TLA+? The answer to that is absolutely! Someone here mentioned Quint, which is also intended for verification, but works much more like a programming language, and is inspired by TLA (the temporal logic in TLA+). Microsoft Research's P programming language (https://p-org.github.io/P/) could be said to be such a language, and I recall seeing several one-person attempts whose names I can't remember. There are also temporal-logic- based programming languages that precede TLA+, like Esterel (https://en.wikipedia.org/wiki/Esterel).

> as soon as you say it realizes unrealizable things, you are speaking poetry rather than programming language design

No! Because the purpose of TLA+ is not to build executable things but to help design and reason about executable things, and it turns out that being able to describe non-executable things is very useful for that purpose (as I tried to show with the Halts example). The ability of a mathematical specification language like TLA+ to describe unrealisable things is the source of its power to succinctly and clearly specify realisable things, because a description of something is not the thing itself.

It's like saying I'm not interested in mathematics, only the subset that can describe physical things. But it turns out that restricting mathematics to physical things makes it more difficult to work with.

This isn't poetry, just the practical realities of mathematics.

> TLC

I think your focus on TLC is a distraction because even when your TLA+ specification does describe a computable process, TLC doesn't actually run that process (it can be put in a mode that does, but that's actually more confusing here). TLC behaves more like a sophisticated type-checker. Type checkers are extremely useful (and TLC even more so) when reasoning about a computational system, and some clever people have found ways to program them to produce interesting outputs, but that's not really what you have in mind when you think about running a program.

For example, TLC can check and verify in an instant a specification of an uncountably infinite number of executions, each of infinite length, and yet choke on a specification of only a few possible instances of, say, QuickSort.

> it seems clear you are talking about a level beyond this, where a TLA+ specification cannot be evaluated even with symbolic tricks.

Yes, but even that is not the point. You seem to be very focused on execution, programming, and evaluation, and these are not the things that TLA+ helps with.

There is no doubt that a programming language is more useful than a mathematical specification for producing software -- because you can build software without a mathematical specification but not without a program -- just as a hammer is more useful than a blueprint when building a cabin, as you can build the cabin without a blueprint, but not without a hammer. But asking how to fashion the blueprint into a hammer misses its point.

Consider this mathematical description of the vertical motion of a projectile thrown from ground level:

    y = v0*t + 0.5gt^2
You can certainly numerically evaluate this formula at different points to create a simulation and plot its motion, and that is very useful, but it's not nearly the only useful thing you can do with the formula. By applying mathematical transformations on the formula (without evaluating it) you can answer questions such as "at what speed should I throw the projectile so that its maximum height exceeds 10m?" or "at what speed should I throw the projectile so that it hits the ground again after exactly 5s?"

The purpose of TLA+ is to write a small specification of the relevant details of a 5MLOC program, and use it to answer questions such as "can network faults lead to a loss of data?" Sure, running the program is the ultimate goal, but being able to answer such questions can be extremely helpful, and is hard to do with a detailed description that's 5 million lines long.

Now, it's perfectly fine to say that you don't care about such capabilities, but these are the capabilities that TLA+ exists to give.

There are languages out there that aim to do both -- produce an executable and offer advanced reasoning capabilities -- but these languages usually turn out to be much more difficult to program with than traditional programming languages and harder to reason with than with TLA+.


> the purpose of TLA+ is to help design and reason about executable things

> execution - [this is not one of] the things that TLA+ helps with.

I think there is a depth limit on HN so I'm just going to stop after this. No, I do not have a "real question". I made a statement, that TLA-PL is a programming language. You still not have agreed or disagreed with this statement, just said that you find it "a distraction" and "confusing" and "not really what you have in mind". Well, un-confuse yourself and present an opinion on its veracity. I don't think it's a distraction because it is a point Lamport brought up in TFA.


> I made a statement, that TLA-PL is a programming language

It is a programming language only the same sense that some subset of English that you could interpret as instructions is a programming language. It's not that you can say, if you only use these symbols then you'd get something executable. While a language could technically be defined like that (a language is just a set of strings), programming languages (and all practical formal languages) are usually defined in a way that it is very easy to mechanically determine whether or not something is in the language and even have a generative grammar for it.

But yes, it is true that are subsets of rich mathematical formalisms (including TLA+) as well as of English that could be "executable", but these subsets are not as easily recognised. So it's a matter of how you define a programming language. If you consider "the subset of English that could be interpreted as a program" to be a programming language, then yes, such subsets of mathematical formalisms including TLA+ exist. If you also require that a programming language is a language that should be easy to decide whether some string is in the language or not, then you'd have to restrict the subset to be very small (just as it is easy to decide the subset of mathematical expressions that can be evaluated by a calculator), and then it's again up to you whether or not you would consider something so rudimentary to be a programming language.

To be more specific, I would not consider the subset of TLA+ that can be simulated or checked by TLC in a way that you'd want an interpreter for a programming language to behave to be a programming language because that subset is not easy to define. For example, you could have two specification of the same program, both of which in the subset that TLC supports, and yet TLC would exhibit a behaviour that could resemble an "execution" for one and not terminate for the other -- even though their meaning is identical. That is because TLC performs a certain analysis of the specification to determine whether some propositions are true or false (e.g. a proposition that a certain variable is always of the same "type") and the algorithm used by that analysis sometimes resembles how people would imagine an interpreter to work and sometimes it doesn't.

So I would say that that subset, like a similar subset of English, is more "a language that could be used to produce a program" than "a programming language" because it is neither as well-defined nor as well-behaved as what we expect from a programming language.

On the other hand, there is an even smaller subset of the language than the one supported by TLC, which could be easily defined, well behaved, and guaranteed to be executable, but it would be quite limited and not make a useful programming language. You can still call it a programming language, but again, it would be smaller than the subset TLC supports.


(Sorry, the + should be a - in the formula above)


P.S.

There's another interesting distinction to be made between Idris, a programming language that can also express a lot of maths, and TLA+, a mathemtatical language that isn't a programming language.

Languages like Idris make a sharp distinction between the type level and the object or computation level. The type level can express something analogous to the propositions you can express with TLA+, but these propositions must be filled or "realised" with content at the object level, and that content is required, even at the syntax level, to be a program, i.e. something that is executable.

In TLA+, in contrast, there's only a "type level". You can define refinement relationships between different specifications -- i.e. one is a more detailed description of the other -- but there is no requirement that any level needs to be executable. It may just so happen that some specification are detailed enough to extract a computation from (same goes for mathematics in general), but that's not a requirement.

BTW, it is interesting to note (as I do here in more detail: https://pron.github.io/posts/tlaplus_part3#algorithms-and-pr...) that quite often even simple algorithms -- I give QuickSort as an example -- that are described in a way that's detailed enough for a human to implement and perhaps even a computer could extract a reasonable computation from, are still not a program. The QuickSort algorithm says nothing about how to pick a pivot because any choice would still be an implementation of QS, even though a program must pick a pivot somehow and some choices may be better than others, nor does it describe in what order to sort the two segments -- they could be sorted in any order or even in parallel. For there to be a program, the computer must ultimately be told these details. Still, the algorithm is specified without them, as they don't matter to the algorithm itself, and it can be formally specified in TLA+ without them. If you choose, you may specify a particular and less abstract implementation of QS in TLA+ and check that it is, indeed, an implementation of the more abstract algorithm.

This could be seen as analogous to the object level in a programming language with dependent types, but the difference is that in TLA+ it's not required nor is there a clear distinction between a level of detail that is executable and one that isn't.


I do think there's an interesting and deep connection between the world of model checking and the world of planning (as most often done in logic languages).


Proving the Riemann hypothesis might also take longer than you live, and in fact may be impossible. So the Riemann hypothesis is not math?


Coq is not a specification language. It’s an interactive theorem prover. The goal set is completely different.


Coq, specifically Gallina, is absolutely a specification tool. It's not only that, but it's one of the big use cases it's explicitly designed to support.


No, it’s not. Gallina is not a specification tool in the way TLA+ is (even if coq calls it its specification language). Gallina is a language used to write mathematical statements which you intend to prove. It’s not designed to write specifications.

Coq is definitely not a specification tool. You can probably prove a specification with it in the same way you actually can do symbolic manipulation with C if you really want to. It still remains an interactive prover.


Of course Coq is also a specification tool. C compilers have been formally verified with Coq. So you have a spec for C in Coq.


"The C standard formalised in Coq" is literally the title of Robbert Krebbers's PhD, that gives you an idea of how usual and easy it was.

The fact that you can formalise a specification in order to prove it doesn't make Coq a specification tool.


Oh, if you define specification as something that should be done by people without a PhD, you might have a point.

I don't think software that needs specs should be done by people without a PhD. Jokes aside, I am not saying that Coq is an easy or simple specification tool. But of course it is a specification tool. Actually, it is one of the very few serious specification tools out there.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: