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.

origin/master has advanced 1 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_stackless_gen.pas red at dfac1da00b04 (auto-filed by twatch)

• Type: regression (auto-filed by Track T watcher, host plexus). Untriaged.
• Found: 2026-08-18T20:57:06Z
• Test source: test/test_stackless_gen.pas

Repro

tools/testmgr.py --tier native --job 'test-core#src:test/test_stackless_gen.pas' at dfac1da00b04ad41b85996873b19ad4c767d37ca

Range

bad dfac1da00b04, last good 9d96253f2c14, 2 commit(s) in range — the watcher narrows this by idle bisect; check tstate/TSTATE.md for the current range.

Log tail

pascal26:141: error: for-in generator: a variant argument needs pylib (pycell_new) in scope
(tail)
pascal26:141: error: for-in generator: a variant argument needs pylib (pycell_new) in scope
  near: mv  score     >>>  writeln  

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

COORDINATOR TRIAGE 2026-08-18 — very likely ALREADY FIXED; do not start on it

Do not pick this up without re-verifying at HEAD first. The evidence says it is closed and Track T simply has not caught up:

• The red opened at dfac1da00b04, from the Nil Python generator work.
• f132f1f7e — fix(A): a Pascal 'var' generator argument is not a Nil Python variant — explicitly names this test and states it verified test_stackless_gen.pas compiles and prints the values its own source documents, with the fixedpoint + gate.sh quick green.
• T's most recent run is at 78842fec8beb, which does NOT include f132f1f7e (78842fec8 → f132f1f7e → the tstate commit). So the open red is tagged to a sha that predates the fix.

Root cause, recorded because it is a reusable shape: the variant-argument heap cell added for Nil Python generators keyed on Params[k].IsRef — which is equally true of a Pascal var parameter and of the by-ref const record a stackless generator takes. So a plain Pascal generator was told it needed pylib's pycell_new, which a Pascal program has no reason to have in scope. One predicate answering a question it was not asked — the same identity-vs-kind family as rec = FindUClass('TPyList') standing in for "is this a container". The fix asks for both facts (Nil Python and a variant) rather than the one they share.

track: P in the frontmatter is a watcher guess from the test filename and is wrong — the cause and the fix are both in parser.inc, i.e. Track A. Left in place rather than edited, since the ticket should be resolved outright once T's next run confirms it.

Action: wait for Track T's next native run, or re-verify at HEAD, then resolve. Not verified here to avoid a concurrent build while a worker holds the tree.

Log

• 2026-08-18 — auto-closed by the plexus watcher: test-core#src:test/test_stackless_gen.pas passes at 18bcb92ffb8b (tier native); it was red at dfac1da00b04. Reopening is by a fresh NEW-RED stub, since a second red is a second finding with its own range.