3 ms·

Rubysyn: Clarifying Ruby's Syntax and Semantics

72 points by petalmind 6mo ago

Smalltalker-80 6mo ago
I prefer the postcard ;-) https://richardeng.medium.com/syntax-on-a-post-card-cb6d85fabf88 https://richardeng.medium.com/syntax-on-a-post-card-cb6d85fa...
- pansa2 6mo ago
  Yes, Smalltalk's syntax fits on a postcard - and it's possible to go even more minimal than that, e.g. Lisp or Forth. OTOH Ruby doesn't need a postcard, it needs a full poster.
- amomchilov 6mo ago
  I never understood the appeal of this talking point. It’s just an accounting trick that moves the complexity to the standard library. For example, SmallTalk is a class based OO system, yet this postcard doesn’t slow you how to create a class.
  - petalmind 6mo ago
    Yeah, I was also going to say that conflating trivial syntax and "supremely easy to learn" is a very weak flex. The first Smalltalk-80 book you randomly find is 560 pages. Also, there is a weird bargaining with "but not block temporaries" and "not an instance variable". Why not? Half of the postcard is basically empty. Also, "except primitives" — how many primitives are there?
    - amomchilov 6mo ago
      Next up: “all the characters you need fit on a single keyboard!”
      - chihuahua 6mo ago
        I hope it's not an APL keyboard https://en.wikipedia.org/wiki/APL_(programming_language)#/media/File:APL-keybd2.svg https://en.wikipedia.org/wiki/APL_(programming_language)#/me...
    - Smalltalker-80 6mo ago
      The appeal of having a very simple, base syntax and almost everything in the library / primitives, would be that code is always easy to read and write, also on the 'reflection' level, where most languages have specific keywords that cannot be used 'first class' with the rest of the language. E.g: Types next to class objects or operator overloading, look at Swift.. (Of course the complexity of a full working system (library) does not disappear in Smalltalk, it's just easier to read and modify. :)
      - petalmind 6mo ago
        > on the 'reflection' level, where most languages have specific keywords that cannot be used 'first class' with the rest of the language. In Ruby this is also the other way around, and it's one of my pet peeves. `Kernel#callcc` looks like a method (arguably, of a special class), yet it's impossible to re-implement directly in the language, as far as I can tell. Same with `raise`, I keep forgetting that it's not a keyword even.
        decasia 6mo ago
        Yeah, I just discovered this and was also amazed. https://decasia.org/tech/2026/02/raise-not-a-reserved-word.html https://decasia.org/tech/2026/02/raise-not-a-reserved-word.h... This being said, I don't think there is any requirement technically that every core method can also be re-implemented in Ruby... There are so many methods that are just thin interfaces to something written in C, whether because they are touching VM internals, or for perf reasons, or because they make system calls or call external libraries.
        petalmind 6mo ago
        You reimplement `while` in terms of `loop`, but do you know that `loop` is also a method :)
SeriousM 6mo ago
> This is an experiment in clarifying some aspects of Ruby syntax and semantics. For that we're going to introduce an alternative Lisp-based syntax for Ruby, preserving Ruby semantics. Lisp? Then I would use Lisp... > The goal is to define a comprehensive, trivially-parsable and sugar-free syntax. Ruby has syntax sugar, no need to remove the funny parts.
- petalmind 6mo ago
  > Lisp? Then I would use Lisp... Lisp has completely different runtime semantics. Even the lexical scope in Ruby is extremely peculiar. One of the hard parts of writing this document was to remove the intuitive influence of Lisp because it just doesn't make sense for Ruby.
- petalmind 6mo ago
  > Ruby has syntax sugar, no need to remove the funny parts. This is just an intermediate representation, it's not meant to be used directly (even though you can do that, of course).
- lifthrasiir 6mo ago
  S-expression is a standard representation for syntax-free semantics. For example, PLT Redex [1] is a DSL for programming language semantics and built on top of Racket which uses S-expressions. [1] https://redex.racket-lang.org/ https://redex.racket-lang.org/
  - petalmind 6mo ago
    Even Ruby `parser` gem uses S-expressions in their documentation: https://github.com/whitequark/parser/blob/master/doc/AST_FORMAT.md https://github.com/whitequark/parser/blob/master/doc/AST_FOR...
  - petalmind 6mo ago
    Do you know if Redex could be a tool that would be useful in my situation? Claude suggests it, but I need to learn a lot of Redex to understand how to apply it. I have a general understanding of what operational semantics is. Or maybe, are there any Redex implementations for common programming languages? \lambda_v is a bit too abstract.

72 points by petalmind 6mo ago

I prefer the postcard ;-) https://richardeng.medium.com/syntax-on-a-post-card-cb6d85fabf88 https://richardeng.medium.com/syntax-on-a-post-card-cb6d85fa...

pansa2 6mo ago
Yes, Smalltalk's syntax fits on a postcard - and it's possible to go even more minimal than that, e.g. Lisp or Forth. OTOH Ruby doesn't need a postcard, it needs a full poster.
amomchilov 6mo ago
I never understood the appeal of this talking point. It’s just an accounting trick that moves the complexity to the standard library. For example, SmallTalk is a class based OO system, yet this postcard doesn’t slow you how to create a class.
- petalmind 6mo ago
  Yeah, I was also going to say that conflating trivial syntax and "supremely easy to learn" is a very weak flex. The first Smalltalk-80 book you randomly find is 560 pages. Also, there is a weird bargaining with "but not block temporaries" and "not an instance variable". Why not? Half of the postcard is basically empty. Also, "except primitives" — how many primitives are there?
  - amomchilov 6mo ago
    Next up: “all the characters you need fit on a single keyboard!”
    - chihuahua 6mo ago
      I hope it's not an APL keyboard https://en.wikipedia.org/wiki/APL_(programming_language)#/media/File:APL-keybd2.svg https://en.wikipedia.org/wiki/APL_(programming_language)#/me...
  - Smalltalker-80 6mo ago
    The appeal of having a very simple, base syntax and almost everything in the library / primitives, would be that code is always easy to read and write, also on the 'reflection' level, where most languages have specific keywords that cannot be used 'first class' with the rest of the language. E.g: Types next to class objects or operator overloading, look at Swift.. (Of course the complexity of a full working system (library) does not disappear in Smalltalk, it's just easier to read and modify. :)
    - petalmind 6mo ago
      > on the 'reflection' level, where most languages have specific keywords that cannot be used 'first class' with the rest of the language. In Ruby this is also the other way around, and it's one of my pet peeves. `Kernel#callcc` looks like a method (arguably, of a special class), yet it's impossible to re-implement directly in the language, as far as I can tell. Same with `raise`, I keep forgetting that it's not a keyword even.
      - decasia 6mo ago
        Yeah, I just discovered this and was also amazed. https://decasia.org/tech/2026/02/raise-not-a-reserved-word.html https://decasia.org/tech/2026/02/raise-not-a-reserved-word.h... This being said, I don't think there is any requirement technically that every core method can also be re-implemented in Ruby... There are so many methods that are just thin interfaces to something written in C, whether because they are touching VM internals, or for perf reasons, or because they make system calls or call external libraries.
        petalmind 6mo ago
        You reimplement `while` in terms of `loop`, but do you know that `loop` is also a method :)

SeriousM 6mo ago

> This is an experiment in clarifying some aspects of Ruby syntax and semantics. For that we're going to introduce an alternative Lisp-based syntax for Ruby, preserving Ruby semantics. Lisp? Then I would use Lisp... > The goal is to define a comprehensive, trivially-parsable and sugar-free syntax. Ruby has syntax sugar, no need to remove the funny parts.

petalmind 6mo ago
> Lisp? Then I would use Lisp... Lisp has completely different runtime semantics. Even the lexical scope in Ruby is extremely peculiar. One of the hard parts of writing this document was to remove the intuitive influence of Lisp because it just doesn't make sense for Ruby.
petalmind 6mo ago
> Ruby has syntax sugar, no need to remove the funny parts. This is just an intermediate representation, it's not meant to be used directly (even though you can do that, of course).
lifthrasiir 6mo ago
S-expression is a standard representation for syntax-free semantics. For example, PLT Redex [1] is a DSL for programming language semantics and built on top of Racket which uses S-expressions. [1] https://redex.racket-lang.org/ https://redex.racket-lang.org/
- petalmind 6mo ago
  Even Ruby `parser` gem uses S-expressions in their documentation: https://github.com/whitequark/parser/blob/master/doc/AST_FORMAT.md https://github.com/whitequark/parser/blob/master/doc/AST_FOR...
- petalmind 6mo ago
  Do you know if Redex could be a tool that would be useful in my situation? Claude suggests it, but I need to learn a lot of Redex to understand how to apply it. I have a general understanding of what operational semantics is. Or maybe, are there any Redex implementations for common programming languages? \lambda_v is a bit too abstract.