fixes

[Byron]

The problem is that with them, we don't notice anymore if the crate changes, because a dependency changes. That also means that older versions of the dependency may stay even though some other crates might pick up a newer version.

Ultimately, this will lead to drift and subtle incompatibilities.

We declare this breaking to enforce a proper re-release.

Note: Had to disable fuzzing again as it fails due to some incompatibility with serde and the nightly compiler that it appears to be using. I really have had it all tonight 😁

[EliahKagan]

Does this change have any bearing on the feasibility of #1510? (I'm asking because I notice that's still open, even though the comment discussion suggests it may not be something that can or should be done.)

[EliahKagan]

It looks like the fuzzing failure may have been temporary, since I ran it in my fork, both on main and at 1757377 from this PR, a few times each, and it ran with no problems:

• https://github.com/EliahKagan/gitoxide/actions/runs/10515699495 (b3ff033)
• https://github.com/EliahKagan/gitoxide/actions/runs/10516173098 (1757377)

Maybe rerunning the failing job here would work as well, and the removal in 3d90ab0 would not be needed.

[Byron]

Does this change have any bearing on the feasibility of #1510? (I'm asking because I notice that's still open, even though the comment discussion suggests it may not be something that can or should be done.)

I think that's unrelated.

Thanks for the hint about fuzzing - I will try again.

Edit: It still fails for me.

[EliahKagan]

Thanks for the hint about fuzzing - I will try again.

Edit: It still fails for me.

That's odd, and I wonder if it is somehow influenced by which event trigger is used (since while I ran it repeatedly, I triggered the workflow via workflow_dispatch from the Actions tab, not via pull_request). I'll see if I can figure anything out.