Repository navigation
build(id): two test failures at main — stale shell-contract v1.3 expectation + sqlx-sqlite back in the lock (blocks code#3720 fold) #4015
Description
Third distinct id breakage today, found by the #3720 fold switch at main e860d3a9f (2026-11-01, switch log on the laptop ~/switch-20261101-145031.log). The package now compiles and the suite runs: 1957 passed, 2 failed (-p id --lib):
1. Stale contract-version expectation
commands::world::tests::the_contract_check_reports_conformance_or_every_violation (src/commands/world.rs:2103):
assertion `left == right` failed
left: "3 frames conform to shell contract v1.4"
right: "3 frames conform to shell contract v1.3"
The report string is formatted from the live contract version (now v1.4); the test's expected string at world.rs:2105 still says v1.3. Looks like a one-line expectation update — but the contract owner should confirm v1.4 is the intended current version rather than an accidental bump.
2. Dependency guard tripped: sqlx-sqlite is back in the lock
store_traits::tests::test_the_lock_has_no_sqlite_with_logsync_wired (src/store_traits.rs:701):
sqlx-sqlite depends on libsqlite3-sys: SQLite is back in the shipped build (see vendor/README.md)
grep -c sqlx-sqlite pkgs/id/Cargo.lock = 3 at main bdc66c7c8. This guard exists to force a decision: either the new dependency that pulled sqlx-sqlite in is unintended (remove / feature-gate it) or the wiring changed deliberately (update the guard with the reasoning recorded). Not fixing this one from the fold side — it needs the dependency owner's call.
Context
Today's sequence for the record: #3787 (voiding_scenarios runtime; fixed by #3871) → #3951 (vt100 dev-dep; fixed on main by d488f321c) → this issue. Each fix revealed the next failure behind it because the suite could not run to completion until now. The fold itself (#3720) is fully merged and staged; a green id suite is the only remaining gate for nixos-rebuild switch on the laptop.
Activity
fix(id): shell contract v1.4 expectation + exclude vendored crates fr…
fix(id): re-resolve the lock without the vendored member graph (#4015)
Merge origin/main into fix/id-4015 (base refresh before landing; no o…
dezren39 commentedon Nov 1, 2026
Fixed in PR #4027 (branch
fix/id-4015):d488f321deliberately bumped the contract revision 3 → 4 with new ScreenPatch frame fields and updated the contract's own tests; only the world.rs expectation was missed. Test now expects v1.4 (the code was right).p2panda-sync'stest_utilsfeature chain (→p2panda-store/test_utils→sqlite) swept sqlx into the lock at the pkgs/id: split the monolith crate into sub-crates along module seams (prioritized by Drew 2026-11-01) #3447 workspace conversion. Fix =[workspace] excludefor both vendored crates +cargo update -w(removals only, zero version changes).Verification so far (laptop):
cargo test -p id --lib1514 passed / 0 failed at the fix tip;id-corelib 238/0; both named tests ok;nix build .#id-webin progress on the final tip — its predecessor run showed the in-sandbox suite at 1987 passed / 0 failed with only the then-unfixed doc tests failing (see #4073).Merge pull request #4027 from developing-today/fix/id-4015
dezren39 commentedon Nov 2, 2026
Merged: PR #4027 → merge commit
ee75087f, verified per the day's protocol (object exists, branch tip is an ancestor of origin/main, content greps in the tree: v1.4 expectation, workspace exclude line, Cargo.lock sqlx-sqlite count = 0).Final verification:
nix build .#id-webon the PR tip EXIT=0 (in-sandbox suite + doc tests green). The id gate for the #3720 fold switch is clear; the switch run follows.