Fix build on NetBSD.

[0-wiz-0]

rustix::fs 1.x has unified the nanonseconds member across operating systems.

Fixes #2004

[0323pin]

@0-wiz-0 thanks for this 👍

[EliahKagan]

Thanks! Per #2004 (comment), I accidentally introduced this bug in #1949 by merging that update PR without adapting gix-index to the changes in rustix.

I have verified that the changes in this PR restore the ability to build on NetBSD, and that I am able to run commands like gix clone with the resulting binary.

A small change, to the commit message only

Although commits that adapt to changes in updated dependencies are most often not conventional (as only some commits need be), in this case the commit is fixing a bug that has appeared in a published version of the gix-index crate.

So I think the commit message, which is currently "Fix build on NetBSD.", will have to start with fix: in order to facilitate proper versioning and changelog entry generation. (It could be anything like "fix: Build on NetBSD." or "fix: Fix build on NetBSD" or some other variation if preferred.)

Since you've checked the box to allow maintainers to push to your branch, I can make such a change and force-push it. I'd be happy to do that, but I figured I'd check first, since you might prefer to do it.

[Byron]

Thanks a lot for contributing the fix, and thanks @EliahKagan for reviewing and validating it!

On another note, now that I read the feedback about the commit message, I just want to share that I have been operating on the implicit rules that

• if I can push into the branch and
• the commit isn't verified/signed

then I can rewrite the commits by rebasing and/or make changes to assure changelog and version generation will be desirable.

Maybe this is something we should make explicit in the contribution guidelines so that in future you can assume consent in that regard.

[0-wiz-0]

@EliahKagan: I've changed the commit message as suggested - feel free to edit it if you want to adapt it even more!

[EliahKagan]

Thanks--looks good!

[0-wiz-0]

Thank you for merging this so quickly!
Do you know when the next release is planned?
We'll have to convince some projects to switch to a fixed version of gitoxide to fix their NetBSD build.

[EliahKagan]

Since there have been no SemVer-breaking changes to gix-index between its most recent v0.40.0 release and here, and this change is non-SemVer-breaking, it seems to me that it may be enough to release gix-index v0.40.1. Projects that specify gix-index v0.40.0 in the usual way should then automatically select gix-index v0.40.1 when built, unless --locked is passed.

@Byron Is this workable?

[Byron]

Thanks for the ping - it's indeed no problem to publish the fix, so it should arrive automatically downstream.

[EliahKagan]

As shown in this gist, I've verified that installation works on the main branch, and when installing from crates.io, since the crate releases associated with the fix here. (That was yesterday, before zip was yanked as described in #2013. This and #2004 are unrelated to #2013, except in that #2013 would make it hard to verify this.)