Stream: git-wasmtime

Topic: wasmtime / issue #12178 windows wasi: rmdir on symlink to...


view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 12:13):

wingo opened issue #12178:

To repro:

    dir.symlink_at(".".to_string(), "dot".to_string()).await.unwrap();
    assert_eq!(
        dir.remove_directory_at("dot".to_string()).await,
        Err(ErrorCode::NotDirectory)
    );

Expected result: fails with NotDirectory

On windows, it succeeds: https://github.com/WebAssembly/wasi-testsuite/pull/198/files#r2630813076

view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 12:13):

wingo added the bug label to Issue #12178.

view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 17:06):

fitzgen added the wasi:tests label to Issue #12178.

view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 17:06):

fitzgen added the wasi:impl label to Issue #12178.

view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 17:06):

fitzgen added the wasi label to Issue #12178.

view this post on Zulip Wasmtime GitHub notifications bot (Dec 18 2025 at 18:40):

alexcrichton commented on issue #12178:

Personally I'd say this is a case where the "spec" for wasi:filesystem should be updated to allow for this case. Windows symlinks are pretty different compared to Unix in subtle ways but there's only so much we can do. I don't think it'd be great if we were forced to, for example, test the file's existence in remove-directory-at first before actually trying to remove it.

view this post on Zulip Wasmtime GitHub notifications bot (Dec 19 2025 at 14:54):

wingo commented on issue #12178:

I keep learning horrible things!!! Agreed that there is no viable workaround. Goddammit.


Last updated: Jan 09 2026 at 13:15 UTC