mirrors/arti.git/crates/tor-cert/src, branch arti-v2.4.0 mirror of https://gitlab.torproject.org/tpo/core/arti http://git.dilluti0n.com/mirrors/arti.git/atom?h=arti-v2.4.0 2026-05-26T10:34:08Z tor-cert: Derive Eq on CertifiedKey 2026-05-26T10:34:08Z Clara Engler cve@cve.cx 2026-05-26T10:34:08Z urn:sha1:a1d98f3d29a63a50f900cb91960dd96def524d5b This commit derives PartialEq and Eq on CertifiedKey and UnrecognizedKey in tor-cert. We will need this later for ntor cross certificates in tor-netdoc. tor-cert: Stabilise everything gated by feature = "encode" (fmt) 2026-04-29T16:49:21Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-28T10:18:38Z urn:sha1:89f0520bb746e70ac78f8857694b2b639b2b2be4 tor-cert: Stabilise everything gated by feature = "encode" 2026-04-29T16:49:21Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-28T10:01:24Z urn:sha1:8c468c37e94e9f0483aba08841aa5ae81ec26a6f This is widely used in-tree already. I don't think it makes sense to feature-gate it. There are some (perhaps rather thin) tests for both the Ed25519Builder and EncodedRsaCrosscert. It is possible we might want to change the API further, but this is still a 0.x crate so that's not going to be a problem. We'll remove the actual cargo feature in the next commit. tor-cert: Ed25519CertBuilder: do builder fn rename 2026-04-29T16:49:21Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-28T09:58:04Z urn:sha1:d90b703d0c4bc4ffb9068fb5c25fc78e72743961 Change all call sites. This completes the rename. tor-cert: Ed25519CertBuilder: do type rename everywhere 2026-04-29T16:49:21Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-28T09:59:15Z urn:sha1:24f82c54e0d6d58544a2434d3ae12c3795f0b3f8 Abolish the type alias and change call references. tor-cert: Ed25519CertBuilder: rename from Ed25519CertConstructor 2026-04-29T16:49:21Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-28T09:49:21Z urn:sha1:cbada95b4f5536eb90cf8b2ba6dc8d0570c017de This is a perfectly ordinary builder type. There isn't any reason why it ought to be called "constructor". And, nowadays, we have things in tor-netdoc called Constructor that take a different approach. Briefly, leave a temporary compat alias, to make diffs more comprehensible. Currently this experimental, so no semver implications. Revert "tor-cert: Provide ed25519 cert decoding via TryFrom trait" 2026-04-27T10:46:52Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-27T10:32:02Z urn:sha1:f7b2aa3d2a09755f64d143cc18df5d5081bd9125 This reverts commit 68caf324a320fff1a4f0e9b0f5014a34d0e3729f. tor-cert: Provide ed25519 cert decoding via tor-bytes 2026-04-27T10:46:44Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-27T10:04:10Z urn:sha1:8a3ff73468aed5f7d26e1cea17f7188bf4c4ed8a This is more sensible and will make the code in tor-netdoc less strange. We'll revert the TryFrom in a moment. tor-cert: Provide ed25519 cert encoding via tor-bytes 2026-04-23T15:45:52Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-23T15:35:33Z urn:sha1:c3080ef6301ce77a1a86a81755961687903d1f89 Implement the Writeable trait. Explain why this approach is correct and leave a comment near the decoder (to avoid future changes making this implementation buggy) and a test case. tor-cert: Provide ed25519 cert decoding via TryFrom trait 2026-04-23T15:45:52Z Ian Jackson ijackson@chiark.greenend.org.uk 2026-04-23T15:35:15Z urn:sha1:f537b840481b4ae00d99bb0be5daa5ca8ab47854 Otherwise we can't implement trait-based decoding in tor-netdoc.