Lightnews — Scholar-powered news

LightNews

Xavier Noria

@fxn.bsky.social

2K followers 140 following 870 posts

Everlasting student · Rails Core · Zeitwerk · Freelance · Life lover

Posts Media Videos Starter Packs

Xavier Noria @fxn.bsky.social · 1h

@evanphx.dev now that I think, I seem to recall you played with some XREAL time ago?

Xavier Noria @fxn.bsky.social · 3h

However, Object was correct.

You had predicates whose name end with a question mark and returned an unspecified object. , You could assume the Object API, but the intention was to used them in boolean context, so to speak.

So `boolish` only less communicative.

Xavier Noria @fxn.bsky.social · 3h

Ah! I used Object before using RBS, I lacked notation.

It is also worth saying my signatures are meant to be just documentation, I don't type check.

Xavier Noria @fxn.bsky.social · 6h

(or non-Pro)

Xavier Noria @fxn.bsky.social · 6h

Has anyone tried the XREAL One Pro for programming? Is it practical?

Xavier Noria @fxn.bsky.social · 10h

RBS Dilemma.

Optional types are marked with a trailing `?`.

E.g., a method that returns `String` or `nil` has a return type of

String?

However, what if the method returns `Object` optionally?

Object

is technically enough, `nil` is an `Object`, but

Object?

may be a didactic redundancy?

Xavier Noria @fxn.bsky.social · 1d

Yes, but now `bool` means `true | false`.

The previous `bool` is now `boolish`.

Xavier Noria @fxn.bsky.social · 1d

See links in the 3rd post.

Xavier Noria @fxn.bsky.social · 1d

Exactly! My predicate signatures had a return type of Object.

Xavier Noria @fxn.bsky.social · 1d

That is my point about the RBS notation.

If the notation tells you that `bool` just means "boolean semantics, no further promise about tbe object", you have it.

Then, if the docs also tell you `true | false` guantee a singleton, you have it.

That is what the wiki had.

Xavier Noria @fxn.bsky.social · 1d

So, you do

raise if denominator.zero?

and the exact object returned by the predicate is irrelevant. You rarely need to commit to return singletons (even if you return them).

And you generally test

assert predicate?

not

assert_same true, predicate?

unless you promise `true`.

Xavier Noria @fxn.bsky.social · 1d

But that is not what true means in Ruby :).

Note that I am writing "true" in regular font, I use `true` to mean the only instance of `TrueClass`.

In Ruby, an object `foo` is true if

p 1 if foo

prints 1.

Xavier Noria @fxn.bsky.social · 1d

In my signatures, the return type for predicates was Object!

Xavier Noria @fxn.bsky.social · 1d

My point is that, idiomatically, that would be `bool`.

The name does not formally restrict the interface, it is what it means, right?

Xavier Noria @fxn.bsky.social · 1d

So if you are told `bool` and that means `top` you know you cannot assume much and the object is mean to be used in boolean contexts.

Xavier Noria @fxn.bsky.social · 1d

you can discuss that with Matz :)

While Ruby is the way it is, all objects are booleans and official stuff should match Ruby senantics

Xavier Noria @fxn.bsky.social · 1d

Not really, in Ruby `nil` and `false` are both as false.

We could also have 0, empty strings, etc. only we don't.

Xavier Noria @fxn.bsky.social · 1d

Links:

Wiki: github.com/ruby/rbs/wik...

Spec: github.com/ruby/rbs/blo...

Xavier Noria @fxn.bsky.social · 1d

I dislike the -ishms because they diminish Ruby semantics.

The symbol `:foo` is not kind-of-true or truish, it *is* true. It is as true as `true`.

And I am not into bending POLS to mean "least surprising to people coming from Java". You do not design a language to accommodate foreign expectations.

Xavier Noria @fxn.bsky.social · 1d

RBS has changed its mind wrt booleans.

The wiki says `bool` is like `top`, since all objects are booleans. Use the union type `true | false` otherwise. And recommends the former as preferable choice. This is the Ruby way.

But the spec defines `bool` and `boolish`. Man, I dislike the -ishms.

Xavier Noria @fxn.bsky.social · 1d

According to the newly shared email, on Sept 18 he was clearly not working for RC anymore.

We have to conclude the password change, okus lack of communication, and all days that lasted are just unnacceptable.

Xavier Noria @fxn.bsky.social · 2d

Media Quality settings!

Xavier Noria @fxn.bsky.social · 2d

"The team of former RubyGems maintainers"

I am seeing this universal quantifier repeated or implied in all this conversation, but I don't think that is correct. Isn't it a subset of them?

Reposted by Xavier Noria

Rachael Wright-Munn @chaelcodes.bsky.social · 2d

Arko told Socket, “I have requested Ruby Central stop claiming to own or create ‘bundler’, although they are welcome to modify the source code and share it under some other name. Ruby Central has replied only to ask for time to review my request with their legal counsel.”

socket.dev/blog/gem-coo...

Reposted by Xavier Noria

Rachael Wright-Munn @chaelcodes.bsky.social · 2d

Maybe from when he trademarked Bundler?
andre.arko.net/2025/09/25/b...
Elsewhere he said that he's asked them to stop distributing Bundler, and they'd asked to review it with their legal team. I can't find that quote though.