Regressions (segfaults) after the latest PR (ranges) · bugs

It seems like my AoC solution for day 1 is now segfaulting after the ranges PR. Should I look into it or is it something you are aware of already?

Matthieu Pizenberg (Dec 05 2025 at 18:17):

print! = |msg| msg.split_on("\n").for_each!(Stdout.line!)

Before the latest PR, this works. But after the ranges PR, I now have to write it again with a for loop to not trigger a segfault. @Richard Feldman I believe you had fixed that issue in another PR a few days back. Some of your changes in the range PR might have re-introduced some negative effect on this line.

Matthieu Pizenberg (Dec 05 2025 at 18:37):

Hum actually it’s not just that. Context matters. I’m looking into it more and trying to make another minimal reproducible example.

Matthieu Pizenberg (Dec 05 2025 at 19:06):

Ok, this is the weirdest bug I’ve ever seen. Here is my minimal example with a segfault on main (4766997).

app [main!] { pf: platform "./platform/main.roc" }

import pf.Stdout

print! : Str => {}
print! = |msg| msg.split_on("\n").for_each!(Stdout.line!)

fnA : Str -> Try(I64, _)
fnA = |_input| {
    var $x = 1
    Ok($x)
}

fnB : Str -> Try(I64, _)
fnB = |_input| {
    var $y = 2
    Ok($y)
}

run! = || {
    print!("A1: ${fnA("test")?.to_str()}")
    print!("A2: ${fnA("test")?.to_str()}")
    print!("A3: ${fnA("test")?.to_str()}")
    Ok({})
}

main! = || {
    _ignore = run!()
}

Matthieu Pizenberg (Dec 05 2025 at 19:09):

❯ roc test/fx/segfault.roc
A1: 1
A2: 1
error: Child process /var/folders/q3/lls0f9bd6x112cdkq8bgjldr0000gn/T/roc-65931/segfault.roc killed by signal: 11
error: Child process crashed with segmentation fault (SIGSEGV)

Anton (Dec 05 2025 at 19:51):

Try getting a stacktrace with gdb: gdb --ex run --ex bt --args roc test/fx/segfault.roc that may get you some extra useful info.

Anton (Dec 05 2025 at 19:55):

I have my OS pretty locked down so that I can safely run codex --yolo or claude --dangerously-skip-permissions. They should be able to find the cause of the segfault in that mode but they don't always find the best fix.

Richard Feldman (Dec 05 2025 at 20:01):

Richard Feldman (Dec 06 2025 at 05:19):

https://github.com/roc-lang/roc/pull/8576 should fix this, although I'm not quite ready to land the PR yet - needs some changes, but if you try that branch it should unblock!

Stream: bugs

Topic: Regressions (segfaults) after the latest PR (ranges)

Matthieu Pizenberg (Dec 05 2025 at 18:00):