Bump payjoin to version 1.0.0-rc.6

[DanGould]

Release the bitcoin_uri public-API insulation along with the pre-1.0
SessionOutcome seal and the proposal txid stability accessor. Bump the
payjoin dependency requirement in the workspace crates and refresh the
minimal and recent lock files. See CHANGELOG.md for the full notice.

(cherry picked from commit 69e5dad4ea3edd985d5b3548b795198c7d68b796)

Disclosure: Co-authored by Claude Code

[coveralls]

Coverage Report for CI Build 30009871065

Coverage remained the same at 86.49%

Details

• Coverage remained the same as the base build.
• Patch coverage: No coverable lines changed in this PR.
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	16196
Covered Lines:	14008
Line Coverage:	86.49%
Coverage Strength:	340.54 hits per line

---

💛 - Coveralls

[chavic]

@DanGould The branch forks from eccaccb, predating #1756, given #1718 and #1747's semantic conflict followed the same pattern, a rebase before merging seems worth the extra CI cycle.

[chavic]

The rc.6 notice says #1747 "seals the persisted SessionOutcome as #[non_exhaustive] and drops its unused witness payload". Rereading the code that landed, the entry describes a different change than the one that was made, in three ways.

1. The enum is not #[non_exhaustive], by design. The derive line has no such attribute (session.rs#L223-L224 (

   rust-payjoin/payjoin/src/core/receive/v2/session.rs

   Lines 223 to 224 in 9e14083

   	#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
   	pub enum SessionOutcome {

   )), and the doc comment directly above it states the opposite intent: "This enum is deliberately exhaustive... A new unhandled variant should error at compile-time" (session.rs#L220-L222 (

   rust-payjoin/payjoin/src/core/receive/v2/session.rs

   Lines 220 to 222 in 9e14083

   	/// This enum is deliberately exhaustive: a terminal outcome is a closed set, and a wildcard arm
   	/// has no meaningful semantics for a caller deciding what a session's conclusion was. A new
   	/// unhandled variant _should_ error at compile-time.

   )). What Pre-1.0 Doorkeeper for SessionOutcome in #1747 actually did is reserve Unrecognized(Txid) ahead of the freeze precisely so the enum could stay exhaustive without a post-1.0 semver break (session.rs#L235-L241 (

   rust-payjoin/payjoin/src/core/receive/v2/session.rs

   Lines 235 to 241 in 9e14083

   	/// The contested outpoints were settled by an unrecognized transaction (neither the
   	/// Payjoin nor the fallback), identified by its txid.
   	///
   	/// NOTE: Nothing in this release produces this variant. It is reserved now because this enum is
   	/// deliberately exhaustive, so a new variant after 1.0 would be a semver-breaking API
   	/// change; naming it before the freeze lets the settlement classifier land additively.
   	Unrecognized(bitcoin::Txid),

   )). A reader following the changelog would add a wildcard arm, which is exactly what the design intends to prevent; payjoin-cli instead added concrete arms (v2/mod.rs#L116 (

   rust-payjoin/payjoin-cli/src/app/v2/mod.rs

   Line 116 in 9e14083

   ReceiverSessionOutcome::Unrecognized(_) =>

   ), #L769 (

   rust-payjoin/payjoin-cli/src/app/v2/mod.rs

   Line 769 in 9e14083

   ReceiveSession::Closed(ReceiverSessionOutcome::Unrecognized(_)) => {

   )).
2. The new variant is not mentioned, but callers must handle it. Since the enum stays exhaustive, Unrecognized(Txid) is an API addition every consumer matching on SessionOutcome has to write code for, even though nothing produces it yet. That seems worth a line in the notice on its own.
3. "Drops its unused witness payload" understates a persisted-format change. Success changed payload from Vec<(ScriptBuf, Witness)> to bitcoin::Txid (Pre-1.0 Doorkeeper for SessionOutcome in #1747 diff), and SessionOutcome is serialized inside SessionEvent::Closed in the event log with plain derived serde, no fallback. An rc.5 log containing a Closed(Success(...)) event will fail to deserialize under rc.6. For a release framed as the last one before the database-compatibility freeze, that consequence seems like the headline of this entry rather than a subordinate clause.

[benalleng]

(cherry picked from commit 69e5dad4ea3edd985d5b3548b795198c7d68b796)

Also this is not a commit that exists here on github. Going to rebase this and update the Changelog to be more precise to the changes that made it into the final commit

[spacebear21]

ACK