Ruby 4.0.0 Preview2

https://www.ruby-lang.org/en/news/2025/11/17/ruby-4-0-0-preview2-released/

by pansa2 • 8 months ago

199 89 8 months ago

ruby-lang.org

lloydatkinson• 8 months ago

I didn't know Ruby has three different JITs

hartator• 8 months ago

Yes, specially ZJIT is news to me.

rurban• 8 months ago

You can ignore it. It's just that the Japanese cannot allow an external jit dominate, and so they came up with a much simplier tier 0 method jit, which will never be able to surpass Maxime's yjit.

shevy-java3• 8 months ago

Mostly the newer ones try to improve the older ones.

I guess at some later point one will dominate and the others will go sleep mode. Just like in the movie Highlander - there can only be one. (I couldn't name offhand which JIT is the main one right now ... I always think it is from Takashi Kokubun but that may now be outdated. MJIT YJIT ZJIT HUJIT WAJIT WTFJIT GRANDMAJIT - too many JITs.)

Lio2• 8 months ago

I was actually hoping it would be four JITs and we'd get Tenderlove's tiny FFI JIT too.

https://railsatscale.com/2025-02-12-tiny-jits-for-a-faster-f...

pjmlp2• 8 months ago

It actually has several more, if you take into account JRuby + all JVM implementations (OpenJDK based distros, OpenJ9, Azul, PTC,...), ART even if not JVM proper with Ruboto, and TruffleRuby on GraalVM.

falcor84• 8 months ago

Any particular changes of interest?

dudeinjapan5• 8 months ago

ZJIT is supposed to be an improvement on YJIT. I'm happy to get any free performance improvements!

The Ruby Ractor (Actor) interface is now completely changed to use a Ractor::Port class, mirroring IPC (inter-process communication) semantics. Ractors were added in 3.0 as a way to get around the GVL/GIL, but having N number of Ruby interpreters running in a Ruby process which would enable executing on N cores at once. For me, hot take but Ractors don't seem to offer major advantages over plain-ol' copy-on-write (COW) forking.

The one "big" feature was supposed to be namespaces, which apparently have now been renamed to Ruby::Box (https://docs.ruby-lang.org/en/master/box_md.html). From what I can glean from the Ruby issue tracker, it appears this feature has been radically descoped, primarily because it had performance impacts, but also, I think probably there are realistic concerns about fit with the existing ecosystem. Unlike Javascript/Python, Ruby has never used "modules" for code isolation--everything is loaded into the global namespace (the "global dumping ground" as I call it.) Now the Box feature is only enabled with an environment variable RUBY_BOX=1

https://bugs.ruby-lang.org/issues/21311

shevy-java3• 8 months ago

I don't think it has really descoped. They just want the transition to be smooth. So they start slowly. The lead japanese dev will most likely in 2026 go for more extension, and I also think this kind of has to be synced a bit with ractor (I think?).

Ractor is also strange. Do many people use ractors? I rarely see them used in actual ruby code out there. Right now it seems to me as if ractors are used by only ... say ... 1% or fewer of the ruby developers out there. A bit more than refinement users ... :P

frou_dh• 8 months ago

I'm happy if that feature has been descoped because when I last read up on it, it seemed like it was being ramrodded in by an inner circle who were hardly interested in constructive criticism even from other Ruby core maintainers.

throw-the-towel• 8 months ago

Isn't everything loaded into the global namespace in Javascript as well? I'm talking about named imports.

_odey• 8 months ago

The NEWS link details a bit more:

https://github.com/ruby/ruby/blob/v4_0_0_preview2/NEWS.md

What is interesting to me is the `Ractor.shareable_proc` changes that solved a bug for a use case I was having. And in general fixes for Ractors make them more appealing to use right now, even though they have not removed the `Experimental` flag from them. They are still missing a built-in concurrency primitive like channels or a lock free queue; I'm curious what they will settle on, Ractor::Port is nice but not enough.

thomascountz• 8 months ago

Ruby::Box (formally: Namespace)

https://docs.ruby-lang.org/en/master/box_md.html

   Ruby Box is designed to provide separated spaces in a Ruby process, to isolate applications and libraries.

shevy-java3• 8 months ago

It is 4.0.0 largely because matz created ruby 30 years ago.

matz is no longer the youngest - although he does look young, he is already 60 years old. He also said he has a retirement plan, e. g. avoiding a situation such as when Guido quit (or semi-quit) from Python (due to fatigue/frustration; Guido is not 100% retired but he is also not necessarily the solo-design-dev either, so it is a bit of a semi-retirement). So we won't know how long matz will be the lead designer of ruby - and who will succeed. Which may be reason to worry depending on who it would be. Imagine DHH takes over - man, there would be an insta-exodus of people ...

So while this release does not have a lot of content as such, one thing that is quite big, even though right now it is not, is Ruby::Box. There are many who don't understand it. The thing is ... I understand the use cases for it. I was not involved in any way with regards to its design, mind you - that was mostly a japanese-group in design. But there are objective use cases for it.

Many years ago I recall on IRC (we oldschool people used IRC back in the pre-discord stone age) some C# hacker said he won't use ruby because there are no strong namespaces, that is, someone else can just overwrite things and then nothing works. Although I think he was a drama guy, and any "danger" to be minimal, objectively he has had a point, simply because ruby had no strong concept of isolation here. Lateron there came refinements. Now refinements are strange, because while I think the use case makes sense, the syntax is strange. Syntax is one huge reason why I do not use refinements; but also because I try to avoid putting my own modifications all over the place, largely because I'd have to distribute that too, and also because modifying core classes, while that has a use, should not be done excessively, IMO.

Ruby::Box kind of builds on that and makes the refinement use case more generic (eventually; I am aware that right now this is not the case but you need a transition stage. Syntax-wise Ruby::Box is also weird, so hopefully the syntax gets easier too, but I instantly understood the use cases. Many people don't, in particular about 95% who demanded a name change away from Namespace to something else, really don't understand the underlying use case.)

Now - making isolated per-project changes is not the only use case. For instance, ractors could be simplified if you know that there are separate ruby processes; ruby threads probably too. These I consider secondary benefits though (and yes, that may be far in the future, who knows; when python removed its GIL though, it put ruby under pressure, aka shape-up-or-go-extinct mode).

One thing I would complain a lot is that on rubygems.org, before RubyCentral went shopify-controlled-only, that people would occupy namespaces. Such as Configuration. I wanted to have a project called Configuration so I can do Configuration.new or Configuration.parse_this_file(). This is possible of course, but when it comes to distributing code, who owns that toplevel namespace? Normally the one who occupied the name first on rubygems.org, sort of. Via Ruby::Box, it should be possible to have ownerships. This could be strong or weak; weak as a hint, aka "psych is owned by ruby core ownershiper but it can be modified", or strong aka making it immutable. Both have use cases. Could also be both. Having this more organized would be really convenient for developers. I would not have to worry whether anyone else uses that "namespace". And of course we need a way to query this state from within ruby code too aka, say immutable:

"If psych is owned by ruby-core, continue to use it."

psych (for yaml) is not a good example here but you can think of any other namespace where you may only want to consider some gems/projects but not others. (Again, the use case may differ between strong and weak ownership, but the thing is that this is an improvement over the prior status quo.)

There are several additional use cases to be had but I'll stop here. What I find strange is that many people who complain, don't refer to the old issues and discussions. We had discussions before refinements were added. About 80% of the people involved, DON'T EVEN KNOW THESE OLD DISCUSSIONS. Either they have dementia, or these are young ruby users who never were active in the old days. It's very strange.

I am not saying all is perfect about Ruby::Box, in particular syntax-wise I'd like improvements, but many people don't seem to understand the use cases, and this is very very strange.

dudeinjapan5• 8 months ago

Ruby::Box in its current form seems to radically miss the mark. Any namespace paradigm only has utility if it is widely adopted across the ecosystem. Javascript did this with ES modules 10 years ago--it IS possible--but it takes a ton of buy-in and effort from the community to achieve it.

I just don't think Ruby has this "burning need" to have namespaces/modules/erm... "boxes". So we're likely to end up with sporadic usage of Boxes leading to inconsistent behavior.

thebigkick• 8 months ago

Curious, why would DHH trigger a mass exodus?

flykespice• 8 months ago

Would you keep using a programming language leaded by a racist? I wouldn't

throwaway489215• 8 months ago

I'd much prefer Jeremy Evans[0] taking over rather than DHH. Jeremy is a Ruby core committer (and much more pleasant person), unlike DHH.

[0] https://github.com/jeremyevans

brigandish6• 8 months ago

The Ruby "community" has long been dogged by politics and culture war stuff. This comment[0] on a flagged submission has lots of links for getting up to speed.

Maybe they're right, there would be an exodus, I just wonder if other languages' communities will want the trouble. The Japanese Rubyists don't care for a second, a nice by-product of Japanese insularity. Ruby would continue in Japan just fine without westerners, don't worry.

[0] https://news.ycombinator.com/item?id=45379652

dymk2• 8 months ago

Ignoring DHH's politics (which are enough to make me look for alternatives), he makes breaking technical decisions to tooling with little to no community involvement. E.g., https://github.com/hotwired/turbo/pull/971

He makes it difficult to trust that, were he in charge of Ruby, he wouldn't just take the reigns and stubbornly do whatever he thinks is technically right for "aesthetics" at the cost of all else.

culi3• 8 months ago

A mantra in the Ruby community is "Matz is nice so we are nice". By contrast, DHH is the type of person who would use the hey.com domain[0] to post rants about London being only 1/3rd white, "Pakistani rape gangs", etc. His racist tirades used to be limited to his twitter, but he's shown he's very willing to use business resources to push his racist politics. More discussions here:

- https://lobste.rs/s/fpri94/dhh_problem_2014

- https://old.reddit.com/r/ruby/comments/1no4zu2

[0] https://world.hey.com/dhh/as-i-remember-london-e7d38e64

werdnapk• 8 months ago

Ruby dev since 2000 and looks like refinements have been available since 2.4. I just used them for the first time this year. Definitely not a go to feature for me, but it was nice the feature was available.

davidw• 8 months ago

As a practical matter, I don't think DHH is a Ruby internals geek, particularly, and seems to have other hobbies as well, so I don't think that's likely to even be a thing the community faces.

ecshafer2• 8 months ago

If Matz ever retires, I hope it is still maintained by Japanese devs, to keep the insanity that is western politics out of it.

bdcravens• 8 months ago

While there's certainly drama of technical and personality in many projects, are there any languages that we can see evidence of "insanity of western politics" being present?

ksec2• 8 months ago

Nothing about Fibres and Async from Samuel Williams ?

Lio2• 8 months ago

Yep having the Async gem in the standard library would be a great win. Async Ruby is fantastic now.

I think I read somewhere that Matz is actually in favour of that but Samuel is holding off for now.

ksec2• 8 months ago

Yes. It should be in 4.0 as experimental at least before being official in 4.1z

kmarc3• 8 months ago

I'm tasked to amend a project written in ruby. With a python background (and some nice pydantic, type annotated, etc "strong" code bases behind my back), every day I spend with ruby is a minefield, a nightmare.

I hoped that ruby4 maybe implements stuff that python has, like type annotations or making the damn parens mandatory, but no. Not surprised that python has ten times more developers according to stackoverflow's survey... I can't possibly imagine a collaborative project where other people also have to work on the same code base, and not having any clue what a symbol under the cursor might be.

No type hints. No mandatory requires. No parens, so never know if something is a method or, callable, or variable. Basically IRB is a must for development, because in the editor, I'm blind.

And the ecosystem is just sad. Swagger-rails libraries out there are rookie jokes compared to what python has. At least there is decent GRPC / protobuf integration, so all new services I am writing can be in python. Or any sane language.

phantasmish2• 8 months ago

I’ve sworn off Rails development for anything short of stupid-high compensation for similar reasons.

Implicit imports (“… which package defined this symbol? Who knows!”), dynamic definitions all over the place (“where’s this defined? Literally nowhere until the program runs!”), all that stuff. It’s awful. I feel blind not being able to answer basic questions about a codebase with grep. And that’s not even considering the lack of static typing.

dalenw• 8 months ago

I used to be a big ruby/rails fan but I have to agree with you. I now write c-sharp and it's a lot less stressful than Ruby. If a Ruby/Rails codebases get to a large enough point it's really difficult to keep track of what types a method you wrote accepts. You end up just constantly double checking your own code. Or you end up with a few type checks and/or type conversions at the top of every method. And maybe I was doing it wrong because it was early on in my career. But when a method can accept literally anything and return literally anything, not even a strong IDE like RubyMine can save you.

chihuahua2• 8 months ago

People on HN seem to hate it whenever someone criticizes Ruby. But the language is a sad joke that's gone on for too long. I totally agree with your points and have many more I could mention (lack of proper debugger support and shitty tooling in general - these things exist, but they break every week)

The 4.0.0 release notes (TFA) are like a joke. Here are the language changes in their entirety:

Language changes: *nil no longer calls nil.to_a

That's it.

zem• 8 months ago

if you just need to write a few hundred lines of code to automate something ad hoc, or play with an idea, or do something exploratory like reverse engineering a file format, ruby is a truly gorgeous language. it just hasn't added the affordances you need to maintain a large code base, which is sad because python has shown that you really don't need that many of them - type annotations and explicit exports that act as namespaces would go a long way.

phantasmish2• 8 months ago

I actually love it for small scripts (not more than a couple hundred lines), including small web services and such. Much more pleasant than Python (though Python handles growing beyond a couple hundred lines way better than Ruby does)

I really like a few of its smaller libraries. Just not Rails.

wiseowise• 8 months ago

What's with the downvotes? Every word here is true.

arp2422• 8 months ago

Because it's just generic complaining on a random Ruby post. It's just boring language flamewar bollocks that we can have on every post on every language.

jazzyjackson2• 8 months ago

Truth <> Relevancy

pkkm• 8 months ago

Similar feelings here. Ruby's philosophy of providing a ton of different ways to do the same thing can lead to some pretty sweet-looking code, almost like poetry... but I'd rather have Python's stylistic consistency and better-integrated type hints. Now that Python has Poetry and uv, Ruby's main remaining advantage has evaporated and it's hard for me to justify using the language.

Another thing I don't like about Ruby is how much the community has embraced the Clean Code brand of readability snake oil. It's easy to come by the opinion that any function over 5 lines is a code smell and over 10 lines it's outright bad. I've even heard the view that if-else statements are a code smell and I should always try to replace them with different classes that have the same interface. To be fair, that only happened twice, but that's two more times than I've heard it from users of any other language. I think that the Python community usually strikes a better balance between avoiding excessive function/class length and avoiding excessive indirection.

throwaway489215• 8 months ago

Another thing I don't like about Ruby is how much the community has embraced the Clean Code brand of readability snake oil.

This also puts me off every time, and I've dealt with people who embrace this a lot as a Ruby programmer. Fortunately, you can ignore them and still enjoy writing Ruby code the way you want.

kmarc3• 8 months ago

I feel you. Rubocop is certainly rather a bikeshedding tool than a linter at this point.

The forced if/else transformations drive me nuts.

arp2422• 8 months ago

This is a classic "I am used to language X. I am now trying language Y, but nothing in language Y is like language X! Why can't language Y be like language X?!"-type post. Yes, Ruby is a different language. It does things differently and priorities different aspects.

dragonwriter2• 8 months ago

I hoped that ruby4 maybe implements stuff that python has, like type annotations

Ruby has had type annotations and typecheckers for quite a while. Unfortunately (IMO) the annotations are not inlinine.

or making the damn parens mandatory, but no.

A linter/formatter can give the effect of that (and tune it for where it is appropraite) whether or not the language has it, though there are good reasons Ruby does not in general.

No parens, so never know if something is a method or, callable, or variable.

Method and local variable are the only possibilities (callables are either the value of local variables or the return value of method calls),

haolez• 8 months ago

I don't know. In my experience, the real nightmare is when a competitor uses Rails and has strong developers. Their velocity is insane and I don't see things falling apart when the project gets bigger. Nowadays, I pray that my competitors use something like .NET or Spring Boot. And this comes from someone who never developed in or lead teams that use Rails.

thiago_fm• 8 months ago

Very little happened in Ruby since 3.2 or even 3, it feels stagnated. Still a very beautiful language from my point of view, but yeah, stagnation.

Wish we'd see more action on RBS, perhaps getting Sorbet to core or something. At least having some sort of consensus in the community.

Or even some considerable project in the JIT, dunno!

Python is definitely ahead in types, work on removing the GIL... list is huge.

honeycrispy2• 8 months ago

I was so disappointed in how RBS was executed it caused me to give up on Ruby and move onto something else. I haven't looked back.

rco8786• 8 months ago

I'm still a Ruby fan but I share your immense, nearly immeasurable disappointment on the implementation of RBS. It's as though someone was tasked to add types to Ruby but do it specifically in a way that would guarantee zero adoption.

multiplegeorges• 8 months ago

You should look into rbs-inline. It's a huge DX improvement. The community spoke up and they are moving in the right direction now.

culi3• 8 months ago

Wish we'd see more action on RBS

This is always my biggest complaint about Ruby. Last time I made a fuss about this on HN someone pointed out rbs-inline which might make it into the core rbhs gem itself soon! This would be huge for the Ruby community imo

https://github.com/soutaro/rbs-inline

pjmlp2• 8 months ago

From where I stand, Ruby has much more interesting work going on JIT land than Python.

Much of the Python stuff, is actually about rewriting Python tools into Rust due to its interpreter slowness and no one caring about PyPy, now apparently there is even a PEP to rewrite CPython!

https://discuss.python.org/t/pre-pep-rust-for-cpython/104906

to-too-two• 8 months ago

With talk here of Ruby stagnating, has anyone checked out Crystal? I have not but have been curious about it.

princevegeta89• 8 months ago

I checked out crystal. It's a nice little language, but it has a very, very long way to go. IDE support is still quite shaky and the developer community is not large enough yet. There are some few good packages and frameworks that work very well though. I think it's still not supported on Windows as of today.

wmoxam• 8 months ago

It has similar syntax, but adding explicit types and macros it's a very different language in practice.

IMO the two languages can be quite complimentary, I've been exploring using Crystal + MRuby to create apps, it's been fun so far.

jazzyjackson2• 8 months ago

One man's stagnation is another's stability, no?

karunamurti• 8 months ago

So Christmas is near the corner then.

Lammy• 8 months ago

I've really loved programming with Ractors in the 3.x series and am excited for Ractor::Port

Ruby 4.0.0 Preview2

Related Stories