Track guessed as P from the test source. The ranker reads frontmatter, so an unset track parks a stub in Track T's queue regardless of what the body says -- correct the track: line if this is wrong.

This expectation records a REFUSAL (a *_fail / {%FAIL} test). Before treating a converged bisect range as an accusation, check whether the named commit IMPLEMENTED the thing being refused -- a feature landing makes its own refusal test go red, and the bisect converges on it correctly. Not a verdict; the tool cannot decide this one.

origin/master has advanced 9 commit(s) since this sha. Re-verify at current HEAD before acting — the callback is tagged to the sha that was tested, which may no longer be the state of the tree.

regression: test-core#src:test/test_default_of_aggregate.pas red at 203438d2cf63 (auto-filed by twatch)

Type: regression (auto-filed by Track T watcher, host plexus). Untriaged.
Found: 2026-08-24T19:22:37Z
Test source: test/test_default_of_aggregate.pas test/test_generic_cycle_fail.pas

Repro

tools/testmgr.py --tier native --job 'test-core#src:test/test_default_of_aggregate.pas' at 203438d2cf6306c543194034cc6980662e5c23bf

Range

bad 203438d2cf63, last good 6d2af7ca7c23, 7 commit(s) in range — the watcher narrows this by idle bisect; check tstate/TSTATE.md for the current range.

Log tail

pascal26:76: error: incompatible types: cannot assign Int64 to AnsiString
pascal26:78: error: incompatible types: cannot assign Int64 to class
(tail)
pascal26:76: error: incompatible types: cannot assign Int64 to AnsiString
pascal26:78: error: incompatible types: cannot assign Int64 to class

Stub ticket: signal only. Track T agent (face 2) enriches or a dev track takes it from the repro line.

Triaged and fixed 2026-08-24 — false positives from the new assignment check

All three reds are the same cause: 497fc8e78 started type-checking assignments (bug-p-an-assignment-is-not-type-checked-at-all), and the rule refused two shapes it should never have seen.

A user-defined IMPLICIT conversion. operator :=(a: TFoo): Integer and Delphi's class operator Implicit(...): TString80 both register under Ord(tkAssign) keyed on the operand type, so i := a and s := t are conversions, not the record-into-scalar store the kind check refuses. The check now stands down when either side has such an overload (AssignHasConversionOperator), and deliberately NOT for OPK_EXPLICIT — an explicit operator is what a CAST invokes, and FPC refuses the bare assignment exactly as this does.
Default(T) lied about its type. For every non-aggregate T the arm built an AN_INT_LIT 0 tagged tyInt64, so s := Default(string) reported "cannot assign Int64 to AnsiString". The tag was wrong before the check existed; the check just made it visible. Zero of a string is '' and zero of a class or pointer is nil, so the arm resolves the type (ParseTypeKind) and builds the node the literal would: c := Default(TCls) is now exactly the node c := nil builds. Aggregates keep their materialised zeroed static.

The 625-pair fpc differential is unchanged by both escape hatches — still 416 accepted, 202 tightenings, 0 regressions — so the diagnostic did not lose any of its reach.

2026-08-24 — resolved, commit 1174c2f54.