Remove extras

[JLesDev]

No description provided.

[MattJackson]

@esheppa — flagging you before we act on this one, since your "update certs generation" commit is on the branch.

This PR (#405, opened 2026-01-13 by @JLesDev) is a large grab-bag. Reviewing it against the current 0.13 work, the substantive parts are either already covered or diverge from the direction we're maintaining:

• Dropping async-std — already done on our branch (feature, dependency, examples/tests removed).
• Clippy/modernization fixes (io::Error::other, div_ceil, u16::MAX, eq_ignore_ascii_case, …) — already applied on our side.
• Removing the native-tls and vendored-openssl backends and flipping the default to rustls — we're intentionally keeping and hardening all three, and not changing the default feature set (downstream users rely on them).
• The xtask CI runner — overlaps with the docker-compose matrix + colima macOS integration lane we already have.

Our inclination is to close it as superseded/divergent, but given your commit is on the branch I wanted your sign-off first. OK to close? If there's an isolated piece worth keeping (e.g. the certs-generation change), happy to pull it out as a focused PR instead.

[esheppa]

Thanks @MattJackson, good to close this.

Regarding some of the changes, perhaps they can be considered separately, if you think it is worthwhile I'll create issues for some.

• TLS libraries. The problem here is that tiberius relies on mutually exclusive features. There are a few ways this could be resolved including by having a TLS trait abstraction within tiberius. Another option which I personally prefer is moving to rustls only, because it is compatible with multiple third party providers including OpenSSL
• Regarding xtask the idea behind this is to allow running the CI logic locally. This is useful when developing, but also useful when changing the CI as otherwise it can be time consuming to run it on actions.

[MattJackson]

Thanks @esheppa, appreciate the quick sign-off — closing #405 on that basis.

On the follow-ups: issues sound like a good way to keep those separate from this grab-bag, so please do open them if you have time — no pressure. On TLS specifically, a rustls-only move is a bigger call than I'd want to make lightly given downstream folks currently lean on the native-tls/openssl backends, but the trait-abstraction angle and the third-party-provider story both seem well worth talking through on their own thread rather than deciding here. The local-CI rationale for xtask makes sense too — as far as I can tell it overlaps with the current compose matrix + colima lane, but there may be an ergonomics gap worth capturing.

For anything worth salvaging from this branch (the certs-generation change included), a focused PR against current main would be very welcome — the 0.13 stack looks to have moved this area along quite a bit. Thanks again.