Allow use of windows-sys 0.61

[djc]

No description provided.

[ctz]

Context: #161

[complexspaces]

Thanks for opening this PR! For some context I tried doing this locally when I did the same for arboard (1Password/arboard#201) a few weeks ago. I found that it was basically impossible to get the version bump to cleanly apply for the sake of CI testing though, so I dropped the idea for the time being.

Due to rustls-native-certs (via schannel) being stuck on an older version of windows-sys with no range flexibility, the cargo update ... command used in CI to force an update to check version compatibility wasn't moving to the right version, and explicitly telling it to bump a revision of windows-sys via windows-sys@... didn't work as another result. You can see that in the relevant Clippy job for this PR its not checking the new version either due to this resolution constriction.

I wasn't confident moving forward with a PR as I didn't have the time to go manually check all the type and function signatures we use for breaking changes.

[complexspaces]

Unrelated to the testing problem but it looks like they finally fixed the definition of CERT_CHAIN_PARA in windows-sys 0.60-0.61, although it doesn't take Windows 7 compatibility into account like our definition does.

[djc]

Fair, I'll prod the schannel folks a bit.

[djc]

schannel 0.1.28 has been released with a bump to windows-sys 0.61. Added a commit that updates the Cargo.lock to include the new stuff.

[complexspaces]

It looks like there were some int -> ptr type changes in the latest windows-sys changes. If you'd like I can try and port my arboard abstraction trait(s) for that over to this repository tonight.

[djc]

It looks like there were some int -> ptr type changes in the latest windows-sys changes. If you'd like I can try and port my arboard abstraction trait(s) for that over to this repository tonight.

Sounds good to me!

[complexspaces]

CI looks happy with the trait EnginePtr type inference shim, so this looks good to go now.

I will refrain from approving since I pushed my own code to the branch though 😄. Feel free to double check it makes sense to you and then consider this comment a 2nd approval for merge if all looks well.

[djc]

Squashed your additional commit and made some stylistic tweaks.

LGTM. ✅

[complexspaces]

I'm super sorry for letting this one wait around, for some reason I thought it had merged already. I've rebased the branch on main to resolve the minor lockfile conflict so it can be merged tonight. No other changes were introduced.

git diff 955e9714aecbfc98219c38a181c6a8d814ea89b5 226141251d9629bd526d836b8436c9645b70c53b:

diff --git a/Cargo.lock b/Cargo.lock
index fa71826..2cf2f4c 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -325,9 +325,9 @@ dependencies = [
 
 [[package]]
 name = "security-framework"
-version = "3.4.0"
+version = "3.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "60b369d18893388b345804dc0007963c99b7d665ae71d275812d828c6f089640"
+checksum = "cc198e42d9b7510827939c9a15f5062a0c913f3371d765977e586d2fe6c16f4a"
 dependencies = [
  "bitflags",
  "core-foundation",
diff --git a/rustls-platform-verifier/Cargo.toml b/rustls-platform-verifier/Cargo.toml
index c34cbb4..bfaee4d 100644
--- a/rustls-platform-verifier/Cargo.toml
+++ b/rustls-platform-verifier/Cargo.toml
@@ -53,8 +53,8 @@ webpki-root-certs = "1"
 [target.'cfg(any(target_vendor = "apple"))'.dependencies]
 core-foundation = "0.10"
 core-foundation-sys = "0.8"
-security-framework = { version = "3", features = ["OSX_10_14"] }
-security-framework-sys = { version = "2.10", features = ["OSX_10_14"] }
+security-framework = "3.5.0"
+security-framework-sys = "2.15"
 
 [target.'cfg(windows)'.dependencies]
 windows-sys = { version = ">=0.52.0, <0.60.0", default-features = false, features = ["Win32_Foundation", "Win32_Security_Cryptography"] }