Stream: ideas

Topic: v0.1.0

Richard Feldman (Dec 20 2024 at 01:59):

I think we're nearing the point where a version 0.1.0 of Roc makes sense. Seems reasonable to aim for 2025 being the year we do it.

Richard Feldman (Dec 20 2024 at 02:00):

(sometime in 2025, not like January!)

Brendan Hansknecht (Dec 20 2024 at 02:00):

Assuming we finish static dispatch, for loops, and like them?

Richard Feldman (Dec 20 2024 at 02:02):

yeah the main things I'd want to consider prerequisites would be:

• lambda sets fixed
• static dispatch landed
• custom types landed
• for and var landed
• CLI changes (e.g. no more roc dev)
• no dylib dependencies in the roc binary other than what the OS requires (e.g. libSystem on macOS and libc on Windows). Linux should ideally be static binaries with musl.
• a way for the CLI to self-update (we've talked about this in the past but I'll start another thread on it)

Richard Feldman (Dec 20 2024 at 02:02):

so extremely definitely not January :laughing:

Richard Feldman (Dec 20 2024 at 02:03):

but those all seem like things that should be plausible in 2025

Richard Feldman (Dec 20 2024 at 02:05):

obviously we won't stop trying to improve the language (0.1 is not 1.0!) but I've been saying we should wait to do an 0.1 release until we have a good reason, and I think a good reason is to communicate "hey we just did a bunch of major changes, but we're going to cool it with those for a bit"

Richard Feldman (Dec 20 2024 at 02:06):

and then that can give us time to try out all the things we've built, get comfortable with them, see what's working as-is and what has room for improvement, etc.

Richard Feldman (Dec 20 2024 at 02:06):

fix bugs, improve performance, clean up code, all that good stuff!

Richard Feldman (Dec 20 2024 at 02:10):

oh maybe I should have put it explicitly in that list, but Windows working too :sweat_smile:

Richard Feldman (Dec 20 2024 at 02:10):

would be a prerequisite

Richard Feldman (Dec 20 2024 at 02:37):

but yeah, I feel like once 0.1.0 is out, it's time to do a promotional blitz

Richard Feldman (Dec 20 2024 at 02:37):

shout from the rooftops about what we've been building

Richard Feldman (Dec 20 2024 at 02:37):

and for everyone to try it out

Richard Feldman (Dec 20 2024 at 02:38):

apparently Primeagen is open to having me on to talk about Roc in ~6 months...he has 655K YouTube subscribers as of today, and who knows how many more by then?

Richard Feldman (Dec 20 2024 at 02:39):

also I can try to do a GOTO Conference about Roc, and their YouTube channel already has over a million subscribers

Richard Feldman (Dec 20 2024 at 02:39):

plenty of ways to get the word out! :smiley:

Anthony Bullard (Dec 20 2024 at 02:48):

Maybe Software Unscripted could do an episode about it. I think they’d have you on

Anthony Bullard (Dec 20 2024 at 02:49):

But I’d love to understand what’s meant by “custom types” here, because I don’t think that word means what I think it means

Richard Feldman (Dec 20 2024 at 02:49):

I'm going to do another episode with José Valim...I was actually thinking about explaining all the static dispatch stuff to him and getting his reaction live

Richard Feldman (Dec 20 2024 at 02:49):

which would also serve to explain it to everyone listening/watching

Richard Feldman (Dec 20 2024 at 02:49):

custom types meaning #ideas > custom types

Anthony Bullard (Dec 20 2024 at 02:50):

I always thought Jose was a lurker here, but I would love to see that episode

Richard Feldman (Dec 20 2024 at 02:55):

I don't think he keeps up with Roc stuff except when he and I chat :big_smile:

Anthony Bullard (Dec 20 2024 at 03:07):

Love the custom types proposal

Anthony Bullard (Dec 20 2024 at 03:08):

So when I finish up with the Docs feedback, I'd like to help on one of these.

Richard Feldman (Dec 20 2024 at 03:27):

that would be amazing! :heart_eyes:

Anthony Bullard (Dec 20 2024 at 03:30):

Looks like there isn't currently an issue for Custom Types - but that seems like one best suited for my level of experience.

Should I create an issue, and maybe we should have either a v0.1.0 milestone or project?

Richard Feldman (Dec 20 2024 at 03:33):

sure, go for it!

Richard Feldman (Dec 20 2024 at 03:33):

I don't have any preferences on milestones/projects/etc. - whatever others agree on is fine by me :big_smile:

Anthony Bullard (Dec 20 2024 at 03:36):

Alright folks who care. Give me :tombstone: for milestones, a :pencil: for project, and :flags: for just a label as a reaction to this message

Brendan Hansknecht (Dec 20 2024 at 05:03):

Exciting stuff

Brendan Hansknecht (Dec 20 2024 at 05:06):

What do we think about the major platform changes:

1. Explicitly passing in allocators and effects as a struct into roc when calling it.
2. Exposing roc type to the host via cffi to make glue manageable and more robust
3. Fixing our mix of cffi issues such that platforms hit less edge cases.

These will be major platform changes. Some breaking. And fixing stuff to make glue functional will be important when we want more platforms. (Especially platforms not managed by luke)

Oh, and I guess if we have any plans for atomics and sharing across threads (but I definitely think that can be post 0.1.0)

Richard Feldman (Dec 20 2024 at 05:09):

great questions! I think all of those wouldn't necessarily be 0.1.0 blockers, unless they turned out to be prerequisites for something else

Richard Feldman (Dec 20 2024 at 05:10):

e.g. it might turn out that for Rust platforms like basic-cli and basic-webserver to work on Windows (which they currently don't, as far as I know) we need to switch to one-way linking, which would be blocked on the allocators change

Richard Feldman (Dec 20 2024 at 05:11):

but the other things I think would be good examples of things to focus on post-0.1.0 while we're getting a feel for how static dispatch and all the other changes feel

Richard Feldman (Dec 20 2024 at 05:11):

because they don't change how it feels to build applications, but rather how it feels to do platform development (as well as their robustness etc.)

Richard Feldman (Dec 20 2024 at 05:12):

also I think atomics and other concurrency-related things should be after 0.1.0, both because concurrency will be a substantial project, and also because it makes sense to have it be its own sort of "news event" if that makes sense

Richard Feldman (Dec 20 2024 at 05:13):

"here is how you can now do awesome concurrency in Roc, check it out!" would be a great blog post

Richard Feldman (Dec 20 2024 at 05:22):

to me, the theme of 0.1.0 is:

• Roc is now reliable enough (after lambda set fixes, Windows fixes, platform fixes) that you can try it out and not worry that the compiler is going to crash all the time. You may still encounter bugs (it's 0.1, not 1.0, etc.) but it's not like it used to be back in the days when I would have huge disclaimers for beginners every time I talked about Roc in public. :big_smile:
• From a design perspective, Roc is now about as friendly to beginners coming from non-FP backgrounds as we anticipate it ever being. So that's purity inference, var, for, and static dispatch. Between this and the previous point, we should get a strong signal from the beginners coming in regarding how easy Roc has become to get into, without the noise of the main blocker being "I keep hitting compiler crashes." :big_smile: Presumably there will still be room for improvement in varioous ways, but this should give us a good sense of the impact the new things have had on the beginner experience!
• Roc is now at a point where you can actually build nontrivial applications (e.g. a realworld backend would be sweet) using things that in 2024 have only been in design documents and discussions, so we can now actually try out the new styles that are available and see how they feel to use in practice!

Luke Boswell (Dec 20 2024 at 06:36):

Anthony Bullard said:

Alright folks who care. Give me :tombstone: for milestones, a :pencil: for project, and :flags: for just a label as a reaction to this message

It's seems pretty unanimous here, milestone inbound

Brendan Hansknecht (Dec 20 2024 at 06:41):

Yeah, your description of 0.1.0 sounds reasonable and matches what I thought you were planning. Frankly a 0.2.0 platform focused release may make a lot of sense. It isn't about the core of the feel of roc, it is about expanding what roc can do and where it can be used. Kinda a way to open up the floodgates of possible applications.

Richard Feldman (Dec 22 2024 at 23:11):

here's a preliminary doc of things that are and aren't blockers - any thoughts on it welcome! :smiley:

https://docs.google.com/document/d/1kA9YfyGSYLD0l2kntH_C28i18hHTECAtz-isWnwAiNY/edit?usp=sharing

Richard Feldman (Dec 22 2024 at 23:11):

(this can be translated into some other source of truth later, I just wanted to get everything down in one place and see if we're missing anything etc.)

Kilian Vounckx (Dec 23 2024 at 09:06):

With roc dev gone, just roc would take its place?

Luke Boswell (Dec 23 2024 at 09:08):

https://github.com/roc-lang/roc/issues/6637 -- cli workflow

Kilian Vounckx (Dec 23 2024 at 09:48):

I see. I have a minor nitpick about just the roc command for scripting, but I'll create another topic

Romain Lepert (Jan 02 2025 at 12:38):

should the spread/rest syntax be included for 0.1.0 ? I suppose it is relevant to the beginner friendliness and the roc language feel.
https://github.com/roc-lang/roc/issues/7091

Richard Feldman (Jan 02 2025 at 12:59):

I don't think it's a blocker, but definitely something we want to do!

Alec Gargett (Jan 22 2026 at 05:59):

Richard Feldman said:

apparently Primeagen is open to having me on to talk about Roc in ~6 months...he has 655K YouTube subscribers as of today, and who knows how many more by then?

He now has 1 million :smile: Maybe when the new compiler and the documentation is done you could reach out to him. He did two videos on Gleam and expressed an interest in having Louis on but that was in response to Gleam v1 and 1.2

Steve Howell (Jan 22 2026 at 17:38):

Richard Feldman said:

apparently Primeagen is open to having me on to talk about Roc in ~6 months...he has 655K YouTube subscribers as of today, and who knows how many more by then?

He's a character. Good luck!

Last updated: Jun 16 2026 at 16:19 UTC