Thread by @zverok

1. A long rant about Ruby, its value, its innovations, and its (sad) fate. Should've probably been a blogpost, but let it be just some random tweets.

2. I do believe that Ruby is an important language, as a milestone/inspiration/step in history. And not because of Rails (despite of it, actually). But what's so important about Ruby?

3. In other words, what's Ruby's selling point, as a language? Or, as @avdi once asked, what's its legacy (avdi.codes/what-is-the-ru…)? →

4. If you try to tell somebody "what's Ruby about", how would that sound? "Well, it is scripting dynamically typed OO language, with a sprinkle of functional and..." and what else? How many languages can be described this way? What's unique, or at least interesting, about Ruby?

(..and that's the reputation it mostly has in the "outside world": just a regular mainstream language, losing popularity. Oh, and "with some weird metaprogramming magic which makes it unreliable", maybe.)

5. I strongly believe that the unique (or at least outstanding) idea behind Ruby exists. Nobody just bothered to name it a "paradigm", to claim the cultural territory. Well, other than the famous "designed for developer's happiness".

6. But "designed for developer's happiness", unfortunately, is not perceived as a substantial claim. From inside the community, it reads as "well, this <gestures broadly at everything> that we love", from outside it reads as "well, that <frowns> that those fanboy hipsters 'love'"

7. Though actually, "designed for developer's happiness" has a substance, and important one, too. In my head, I tend to call it "phrase-oriented programming": the language is thoroughly designed the way that each particular phrase would be clear to write and read.

8. That's actually hard to reason about, without appeal to "feelings", but Ruby's novelty lies exactly here: allow to write the "phrase" (expression) describing complicated concepts, with every element meaningful, necessary, and unambiguous.

9. Obviously, I don't know what exactly @yukihiro_matz thought on all design stages, but my guess is it all started with Enumerable: the amount of "design elements" which made `for` obsolete is surprisingly low, but their mix is "just right".

10. Ruby actually introduced a lot of small (phrase-level!) innovations like this. I want to remind you that at the moment of the Ruby unexpected ascend, passing around functions in most of mainstream languages was as tedious as map(function(x) { return x**x; })

(Maybe it is a stretch, but I believe that "shorter" lambdas like map(x => x**x), or Swift "tail lambdas", or Rust blocks, were largely brought forward of Ruby alumni or at least Ruby-curious people.)

11. I would go so far as to suspect that the classic OO part was somewhat "accidental": that's what was mainstream at that point, and Ruby was pronouncedly non-esoteric language: just your regular class/method/if/return/math, but with critical attention to the phrase level.

12. Phrase-level expressiveness requires clarity of each element. In Ruby (almost) everything is an object or its method, and `self` is always implied and clear. Syntax variability is not a toy: it allows lower the number of concepts, so we don't need separate "attributes".

13. ...and most of what looks like keywords can also successfully be method calls: like attr_accessor, private, raise (but Ruby is demonstratively non-esoteric, so it doesn't sacrifice `if`s, like Smalltalk did).

14. ...and whatever you need is typically a method of the current context or object you want to do something with, discoverable and guessable (I still remember my post-C++ fascination when I first blindly tried 1.3.floor and it just worked as expected)

15. Low amount of base concepts + rich dictionary + flexibility of syntax + design of core classes might make you think about problems in a completely new way. "Phrase-based" or "phrase-oriented", as I've said.

16. It goes like "I'd _like_ the final solution of this problem to be _spelled_ this way. It is a correct Ruby code. So let's design a few classes/objects/methods so this would become possible". And repeat, all the way in those classes/objects.

(One might notice that it resembles some of BDD concepts! And for a reason. I'll talk about it down the thread.)

17. Well, Ruby's initial design was not ideal (as somebody working with it since Ruby 1.6 I can tell). The Module ambiguity (mixin/namespace), text-based `require`s, weird convention of `Array()` and alike method-like-classes... But most of it hit a very sweet spot.

18. But what Ruby lacked was a "killer app": some critical mass of code that would demonstrate how it allows thinking differently. For me, this "killer app" was the language itself: I always wanted to design my code the way Ruby's core designed.

19. But typically that's not enough. And even Ruby's core library was not a good candidate for the showcase (Net::HTTP anyone?..)

20. OK, by this point everybody knows what we are talking about next: Rails!
And here we come.

Let's put it simply: Rails is not, and never was, the "showcase killer-app of Ruby's coolness".

21. Why is it so?
Ruby's unique proposal: find the way your ideas is expressed the best way, design your dictionary, combine it in different ways, see where the combinations lead you. It is the proposal of a modelling clay: you have it as a material, and go wherever you want.

22. Rails' unique proposal: somebody took the modelling clay, created a language in a shape that was natural to them, and gave it to you as a ready-made (baked) _frame_ with small holes you now can fill.

23. When somebody says "but Rails popularized Ruby!", the only thing I can answer: no, it did not. The way for Rails to popularize Ruby was by saying "Look, we described in Ruby how we saw _our_ app! Let it be an example of _any_ design describable naturally"

24. As we all know, that's not what happened. I have huge respect for Rails' initial authors as people, businesspeople, and thinkers, but what happened with Ruby being "popularized" by Rails is downright tragic.

25. Another "tragedy", of a smaller scale, happened with RSpec. RSpec was—still is!—a much better "killer app" for Ruby's expressiveness; the whole bunch of ideas for _one-line specification of expected behavior_, combinable with any third-party matchers is aligned with Ruby way.

26. But slowly-yet-steadily RSpec's preferred ways (even by its current maintainers—counting from the removal of its() statement) shifted from phrase-level expressiveness to "frameworkness", which should be filled with "simple" code.

(I ranted about it some time ago: zverok.github.io/blog/2017-11-0…)

27. It is perfectly understandable that most of us who stayed are primarily maintaining Rails codebases (or—codebases in alternative frameworks, with the same "frameworky" approach, but allegedly better designed).

28. Obviously in those conditions "phrase-level expressiveness" (especially it doesn't even have a catchy industry-approved moniker!) is long out of focus of the community.

29. New language features towards expressiveness are now met with hostility or indifference. "Ruby-the-language good enough for me, now I want those tests to run 10% faster". We are all pragmatic product engineers, no offense.

(I still believe—and, kinda, paid for doing so—that short-and-expressive code has a business value, and the time spent trying to understand "how to say those 30 lines in one obvious phrase" is not a time wasted, but that's a personal trait, not a merit.)

30. What makes me sad is that the usages of Ruby where it could've shine, are mostly out of question now. I could be a perfect modelling clay for teaching, experimentation, trying new areas. Let those clay models, once proven good, be repeated in more steely languages, so what?

31. But alas, Ruby as a tool for thinking new thoughts is mostly off the road of IT culture and history. It is hard to bring "in Ruby, we discovered this! wanna try too?" into the general discussion in 2021.

32. Sometimes I wish that the Rails "hype" died to the ground, and for a few years, people stopped to try building water towers of modelling clay, and let it became half-forgotten and then, once, rediscovered for its true value. But that's just old and grumpy me.