← board

Track T by default: the FAILING STEP named no owner. Line 2 of 2 is tools/expect_same.sh sweep_fieldrooted_frozen_default "$(/tmp/sweep_fieldrooted_frozen_d)" "$(cat test/test_field_rooted. The job's own src (test/test_field_rooted_nested_dyn_frozen_index.pas, 3 file(s)) is NOT used here on purpose: it is what the job compiles, not what broke, and guessing a lane from it is what sent three reds in one job to the wrong lane. This is a FALLBACK, not a finding — nothing says the defect is Track T's. Re-lane it before working it.

origin/master has advanced 2 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_field_rooted_nested_dyn_frozen_index.pas@1 at 91b4b77ec631 in step 2/2, tools/expect_same.sh sweep_fieldrooted_frozen_default "$(/tmp/sweep_fieldrooted_frozen_d)" "$(cat test/test_field_roote… (auto-filed by twatch)

Repro

tools/testmgr.py --tier native --job 'test-core#src:test/test_field_rooted_nested_dyn_frozen_index.pas@1' at 91b4b77ec631e4027893233277450f576e3008fc

Range

The named sha 91b4b77ec631 CANNOT be the cause — it touches no buildable file (docs / tickets / tstate only). It is the sha that was TESTED, i.e. the upper bound of an untested range; the cause is somewhere below it.

bad 91b4b77ec631, last good 5e2dcc37c253, 1 commit(s) in range — the watcher narrows this by idle bisect; check tstate/TSTATE.md for the current range.

Log tail

Segmentation fault (core dumped)
(tail)
ok: /tmp/testmgr-scratch-1627775/sweep_fieldrooted_frozen_d  [code=69400B  data=3544B  bss=43524B  procs=134]
Segmentation fault (core dumped)
expect_same: MISMATCH [sweep_fieldrooted_frozen_default]
--- expected
+++ actual
@@ -1,5 +1 @@
-VAR   <var0><var1>
-FIELD <fld0><fld1>
-LEN   2 1
-ANSI  <ansi0><ansi1>
-INT   41 42
+

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

RE-LANED T/P -> A by the coordinator, 2026-09-03, on COMMIT CONTENT, not on the failing step. The only code commit between the last GREEN native (5e2dcc37c253) and this RED (91b4b77ec631) is 0dedfb86c "fix(A): SetLength through a field, an element or a deref", which touches Makefile, compiler/ir_codegen.inc and the 386/aarch64/arm32/riscv32/xtensa arms plus compiler/pasparser_stmt.inc. The other five commits in the window touch no code at all. That is a suspect established by ref arithmetic with zero builds — it is NOT a measured cause, and the author (franka-29) has been told and may reject it.

This may not be a regression. The alias-cast row fails on a REFUSAL grep (SetLength expects a), and 0dedfb86c deliberately widened what SetLength accepts. A refusal test going red is exactly what an intended widening looks like; if that is what happened, the TEST is what needs updating and there is no defect here. Establish which before treating any of these three as a bug.