Tracking a coverage gap I introduced while rebuilding #140, so it is not lost in a PR comment.
What happened
#140 shipped with committed merge conflict markers in sdk/tests/authored-flow.test.ts (12 of them, citing 0987e38), so the branch did not compile. Resolving them mechanically was not possible — four of the six regions have brace_delta=2, paren_delta=1 on the incoming side: unbalanced fragments whose closing braces live in shared trailing context. The two sides interleave.
I took main's version of the file whole. That guarantees main's merged lifecycle and refusal coverage survives — the property worth protecting, since silently dropping shipped tests is a failure this stack has hit repeatedly — but it means four cases the branch had added are not in 066ef24..:
- passes direct input into the journal-backed authored body
- preserves separately awaited sibling ordering before the join
- requires an explicit completion after journal-backed steps
- an
it.each([...]) table (contents not recoverable from the conflict hunks alone)
Plus a connectedClient(name) helper those cases used.
Why this is a gap and not a crisis
The direct-input feature itself is still covered: sdk/tests/direct-input.test.ts is a separate file and passes. What is missing is coverage of direct input interacting with the authored-operation lifecycle — ordering around the join, and completion requirements after journal-backed steps.
So this is a real reduction in interaction coverage, not an untested feature.
What is needed
Someone who knows the intended bodies should re-add them on top of current main. The originals are recoverable from git show 6384600:sdk/tests/authored-flow.test.ts, but the incoming halves are fragments — they need reconstructing, not copying.
Raised by the session that rebuilt #140. I did the removal, so I am flagging rather than quietly absorbing it.
Tracking a coverage gap I introduced while rebuilding #140, so it is not lost in a PR comment.
What happened
#140 shipped with committed merge conflict markers in
sdk/tests/authored-flow.test.ts(12 of them, citing0987e38), so the branch did not compile. Resolving them mechanically was not possible — four of the six regions havebrace_delta=2, paren_delta=1on the incoming side: unbalanced fragments whose closing braces live in shared trailing context. The two sides interleave.I took
main's version of the file whole. That guarantees main's merged lifecycle and refusal coverage survives — the property worth protecting, since silently dropping shipped tests is a failure this stack has hit repeatedly — but it means four cases the branch had added are not in066ef24..:it.each([...])table (contents not recoverable from the conflict hunks alone)Plus a
connectedClient(name)helper those cases used.Why this is a gap and not a crisis
The direct-input feature itself is still covered:
sdk/tests/direct-input.test.tsis a separate file and passes. What is missing is coverage of direct input interacting with the authored-operation lifecycle — ordering around the join, and completion requirements after journal-backed steps.So this is a real reduction in interaction coverage, not an untested feature.
What is needed
Someone who knows the intended bodies should re-add them on top of current
main. The originals are recoverable fromgit show 6384600:sdk/tests/authored-flow.test.ts, but the incoming halves are fragments — they need reconstructing, not copying.Raised by the session that rebuilt #140. I did the removal, so I am flagging rather than quietly absorbing it.