cc @robertbastian what's the status of icuexportdata being updated? I thought we were already on Unicode 16.

All I can tell you is that 2.0.0-beta1 is on ICU release-76-1. Whether that correctly exports Unicode 16 requires me to debug ICU4C, which I'm not familiar with.

It's supposed to , from the relnotes: https://github.com/unicode-org/icu/releases/tag/release-76-1

Confirmed that this reproduces on ICU4X main, and confirmed that Unicode 16 data has a whole bunch of scx values for low codepoints that are not available on Unicode 15.

Trying to build ICU4C to see what's up

Found the culprit: https://unicode-org.atlassian.net/browse/ICU-21821

That hardcoded table in icuexportdata needs to be updated

cc @sffc @echeran

New data in #6044

Confirmed that it passes the following test:

#[test]
fn expected_script_thing() {
    use crate::props::Script;
    use crate::script::ScriptWithExtensions;
    let scripts = ScriptWithExtensions::new()
        .get_script_extensions_val('\u{2bc}')
        .iter()
        .collect::<Vec<_>>();
    assert_eq!(
        scripts,
        [
            Script::Bengali,
            Script::Cyrillic,
            Script::Devanagari,
            Script::Latin,
            Script::Thai,
            Script::Lisu,
            Script::Toto
        ]
    );
}

Linking #4602

It seems you have a workaround for this for now: We have fixed the ICU4C data export around this, and could do the work for a patch release, but @sffc and I would prefer to wait till ICU4X 2.0.0-beta2 which should happen in the next few weeks, instead of doing a transient patch release.

A few weeks is fine, thanks for tracking this down!