Stream: contributing

Topic: Google summer of code

Hubert Małkowski (Jan 23 2026 at 13:42):

Hey, will Roc participate in Google summer of code this year? I'd love to participate :)

Richard Feldman (Jan 23 2026 at 22:28):

I don't have any current plans, but one project idea I had in mind is creating a pure-Roc numeric computing library and benchmarking it against industry standards like BLAS

Richard Feldman (Jan 23 2026 at 22:30):

wouldn't need to be feature-complete, mainly would be about learning whether we can be competitive on performance without resorting to FFI, which is very relevant there bc all the math operations are pure functions, but in Roc FFI is done via effectful functions

Richard Feldman (Jan 23 2026 at 22:31):

so it would be really valuable to see if we can get competitive perf in those use cases in pure Roc, because it keeps coming up that people new to the community want to use Roc for use cases where the usual move would be to get BLAS (or similar) via FFI, and the discussion always gets stuck on hypothesizing whether pure-Roc would actually work in practice from a performance perspective :smile:

Richard Feldman (Jan 23 2026 at 22:32):

is that a project that would interest you @Hubert Małkowski?

Luke Boswell (Jan 23 2026 at 23:10):

Do GSoC contributors work full-time or part-time on their projects, and are they getting paid or is it all voluntary?

Hubert Małkowski (Jan 24 2026 at 10:37):

Richard Feldman said:

is that a project that would interest you Hubert Małkowski?

It very much would :)

Luke Boswell said:

Do GSoC contributors work full-time or part-time on their projects, and are they getting paid or is it all voluntary?

Part-time in a span of a few months (typically June to September). Paid by google if project succeeds.

Richard Feldman (Jan 24 2026 at 12:56):

ok cool, I can get an application going as an org!

Hubert Małkowski (Jan 24 2026 at 13:09):

omg, that's awesome! Thank you!

Matthieu Pizenberg (Jan 25 2026 at 18:21):

Niiiice. Let’s get some numbers for how fast pure roc math can go :)

Blas and lapack are huge libraries, optimized for years so I wouldn’t be surprised it’s hard to beat considering how much impact is also due to memory organization, vectorization, and we don’t have control at all in Roc for these.

Steve Howell (Jan 29 2026 at 14:29):

I have worked on the Zulip project while it's been one of the most active GSOC projects since ~2017. I can tell you that it's a LOT of work to inbound the students. But the positive side is that you WILL get long-term contributors from it. And maybe even a trip to California (I went to the Google campus one year as part of the GSOC gig). https://blog.zulip.com/2025/12/02/google-summer-of-code-2025/ is interesting.

Last updated: Feb 20 2026 at 12:27 UTC