Ah, the joys of having a compatible data model! I think it’s really cool that Gleam and Elixir can interop so well because they’re both running on the BEAM and can share data structures.
I’m curious to see how Elixir’s very recent (still prototypical) type system addition will change the demand for a language like Gleam.
I don’t think that Elixir’s types and Gleam’s have much in common, they will provide very different experience. Coupled with there not being much crossover from the Elixir community to the Gleam one I don’t think it’ll have much impact on Gleam.
First, let me make it clear: I'm not trying to knock Gleam and it's type system—it's really cool what you've done.
> Not much crossover from the Elixir community
Interesting! I didn't know that. For the die-hard ML-style type system fans (of which I am one) Gleam will absolutely be the most attractive thing on the BEAM block. I think there is a small contingent of developers who would really rather have a type system who might choose Gleam, but that Elixir's new type system (when it matures to support user-supplied annotations and whatnot) might suffice. I think it's an interesting case to see how adding a type system might shift people's preferences between two established languages.
That said, again please don't see this as a knock on Gleam or me trying to say "well now Gleam is obsolete"—I'm not trying to say that. I hope Gleam has an illustrious future. :)
I thought would be the case too originally, but now that Gleam has been known long enough to get an idea for the sort of people who are attracted to it I don't see Elixir getting types having any real impact on Gleam's uptake.
Elixir and Gleam are very different languages offering very different things.
Elixir's type system is still baking(can't define customs types yet), while Gleam's is ready here. Gleam has (familiar to many) battle-tested Hindley-Milner with great type inference, while Elixir's is a different (still sound) set-theoretic one.
But most importantly, Elixir is, and always will be, a gradually typed language. There will always be parts of it that will remain or encourage dynamic typing and that opens the door to ambiguity. In a statically typed language the compiler enforces type annotations and that is where Gleam shines.
Sure sure, I don't disagree with anything you've said. I do think there are some people who would really have some kind of type system than none at all, and for whom the existence of Gleam might be enough for them to stay away from Elixir; with a (gradual) type system though, that might be enough for a few of them to come back. I'm just curious to see how the landscape shifts.
I’m curious to see how Elixir’s very recent (still prototypical) type system addition will change the demand for a language like Gleam.