Tracking: minor follow-ups named in the #93 review (post-merge) #97

PR #93 merged on 2026-10-06 — thank you again for the review and the merge. The blocker and majors 1–6 were fixed and gated before merge (fix summary). Opening this small tracker so the minor findings named inline in the review comment don't get lost post-merge.

Line numbers below refer to the reviewed revision (ff1e47a); they have shifted since (majors 3–6 touched the same files).

• ;-split false rejection (gql_transformer.cpp:1320 region) — input containing ; can be wrongly rejected at the statement-splitting boundary.
• ACYCLIC (n) edge case (gql_transformer.cpp:2133 region) — edge case in the ACYCLIC (n) path-mode handling.
• fixed-{n} up to 1000 bypassing width caps (gql_transformer.cpp:906,929 region) — quantified path pattern expansion.
• Recursive JSON descent depth — nesting depth in the JSON value bridges is not bounded.
• jsonEscape control characters — control characters in string values when producing JSON.
• addFunc silent shadowing — registering a function over an existing name does not warn.
• Note-tier weakness — the 120 "passed-with-note" TCK results are weak evidence for "green"; worth considering a stricter classification in the report.

Notes:

• The full minor list ("available on request") was never posted before the merge; this tracks the seven items enumerated in the review comment itself. Happy to fold in more if the rest surfaces.
• Triage rule on our side: anything that can surface a wrong result silently (rather than throwing or being documented as a known gap) jumps the queue — our first reading is that addFunc shadowing and jsonEscape control chars are the two candidates for that class, but each item gets verified before any fix.
• Happy to send these as small follow-up PRs here — or to take direction on which to prioritize.