Created attachment 896118 [details] build.log and emerge --info Unfortunately NOT a strictly big-endian issue, as it passes on ppc64. Reproducible. failures: ---- build_context::test::test_macosx_deployment_target stdout ---- thread 'build_context::test::test_macosx_deployment_target' panicked at src/build_context.rs:1275:9: assertion failed: `(left == right)` Diff < left / right > : ( ( 10, < 7, > 12, ), ( 11, 0, ), ) note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace failures: build_context::test::test_macosx_deployment_target test result: FAILED. 44 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 7.64s error: test failed, to rerun pass `--lib`
Any idea/hint about what this may be about? (I know next to nothing when it comes to sparc). Does old maturin-1.4.0 still pass? I believe the tests passed on catbus before. Could also try the new maturin-1.6.0 and skip stable for 1.5.1-r1 if it helps. Not that the test itself seems very interesting for us so i'd see little harm in skipping it, we mostly care about building packages using the pep517 backend and I don't see ::test_macosx_deployment_target affecting that. Could try building a random package that uses maturin just to be sure it works.
(In reply to Ionen Wolkens from comment #1) > Any idea/hint about what this may be about? (I know next to nothing when it > comes to sparc). > > Does old maturin-1.4.0 still pass? I believe the tests passed on catbus > before. Retried 1.4.0 on amd64 to be sure it still works somewhere and that the test existed: test build_context::test::test_macosx_deployment_target ... ok If 1.4.0 fails, guess it'll either be something different with catbus, or something that went wrong with newer rust (wouldn't be too surprising on sparc). Not much else I think of as suspect. If it passes, then guess it's going to be some new crate.
I'll try 1.4.0.
(In reply to Sam James from comment #3) > I'll try 1.4.0. 1.4.0 fails using rust-1.79.0. If only Rust were slotted...
rust-bin-1.78.0 fails
rust-bin-1.77.1 fails
rust-bin-1.76.0 fails
rust-bin-1.75.0 fails...
rust-bin-1.74.1 fails, not sure if it's worth going back further?
(In reply to Sam James from comment #9) > rust-bin-1.74.1 fails, not sure if it's worth going back further? No, I don't think so. Let's just take it to upstream at this point and ask. I noticed there are rustc version checks nearby to the failing assertion which makes me nervous.
tests pass (!) with 1.73.0
(In reply to Sam James from comment #11) > tests pass (!) with 1.73.0 wow, thanks for going back that far -- guess that may potentially be an issue in rust itself then Well, as far as downstream goes, I don't think this need to block stable (1.4.0 is no better) -- so guess I'll skip the test on sparc.
(In reply to Ionen Wolkens from comment #12) > (In reply to Sam James from comment #11) > > tests pass (!) with 1.73.0 > wow, thanks for going back that far -- guess that may potentially be an > issue in rust itself then Then again, if there are rust version checks, who knows...
(In reply to Ionen Wolkens from comment #13) > > guess that may potentially be an issue in rust itself then > Then again, if there are rust version checks, who knows... Yeah, I just (now) seen the discussion in #gentoo-python -- reposting the link here for reference: https://github.com/PyO3/maturin/blob/main/src/build_context.rs#L1279-L1280 So I don't suspect rust too much either now.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=6f4e16b5e58cc8fefc98b02ca05f9f8e067effa5 commit 6f4e16b5e58cc8fefc98b02ca05f9f8e067effa5 Author: Ionen Wolkens <ionen@gentoo.org> AuthorDate: 2024-07-01 00:10:17 +0000 Commit: Ionen Wolkens <ionen@gentoo.org> CommitDate: 2024-07-01 00:32:30 +0000 dev-util/maturin: skip failing test_macosx_deployment_target on sparc Been failing since rust-1.74 likely due to how it does something different for >=1.74, and then there is no reason for this to block stable given 1.4.0 is affected as well. Plus that a macosx deploy test fails seems unlikely to affect building revdeps with the pep517 backend (seems unlikely anyone uses it for more than that on sparc). Bug: https://bugs.gentoo.org/934573 Signed-off-by: Ionen Wolkens <ionen@gentoo.org> dev-util/maturin/maturin-1.6.0.ebuild | 3 +++ 1 file changed, 3 insertions(+)
