Repository navigation
oracle: zecs-zecs001 case with two stated ZUIDSTCK TODs; TEST AFTER VARYING and short-record fixes (Part of #5945) - #5979
Conversation
…Part of #5945) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ARYING and short-record padding fixed (Part of #5945) The case posts to the zECS cache with two stated ZUIDSTCK TODs; both runtimes agree on every event and on the key record. Two engine defects surfaced and are fixed: PERFORM WITH TEST AFTER VARYING was emitted test-before, and a WRITE ... LENGTH(n) short of the record was not padded with spaces in the entity's fromRecord. The harness resolves install-symbol templates in files.cfg names. Register X47 records the evidence level. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
No description provided. |
|
No description provided. |
|
No description provided. |
|
PR #5979 @ be4861a: 2 failed check(s) == ground-truth (failure) -- real == ground-truth part 4/4 (failure) -- real |
|
No description provided. |
…ime catalogue (Part of #5945) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
CI shepherd: the head of this PR (f9fcefb) is not the one a maintainer labelled |
|
No description provided. |
|
No description provided. |
|
PR #5979 @ f9fcefb: 2 failed check(s) == db2 4/6 (failure) -- infra == det (failure) -- real |
|
CI shepherd: the head of this PR (60aed3c) is not the one a maintainer labelled |
|
CI shepherd: the head of this PR (60aed3c) is not the one a maintainer labelled |
|
CI shepherd: the head of this PR (6049122) is not the one a maintainer labelled |
|
No description provided. |
|
No description provided. |
|
No description provided. |
Part of #5945 (slice 2 of 3; slice 1 is #5968).
What changed
tests/equivalence/le/zuidstck.c: the oracle twin. ReadsGGC_ZUIDSTCK_TOD(32 hex digits), writes 16 bytes, exits 98 by name when unstated or malformed.Zuidstck.javagainsstate(hex)(all scenarios share a JVM).tests/tools/equivalence_cics.py:todcase key (per scenario, both runtimes),ZUIDSTCKin thecalls_leregex, and install-symbol templates resolved in files.cfg names.tests/equivalence/zecs-zecs001: POST to the zECS cache with two stated TODs.det_port.py run: 561/561 translated, PROVED; the 6-byte key prefix is the stated TOD's on both runtimes.PERFORM ... WITH TEST AFTER VARYINGwas emitted test-before (det/gen.py); aWRITE ... LENGTH(n)short of the record was not space-padded in the entityfromRecord(cobol_to_java_repository_forge.py).Evidence
Translated 561/561 (the COMMAREA DTO hole from the earlier count is not hit by this case); executed-equivalent on 2 scenarios,
application/octet-streampath only (atext/*body hits the stub's second-WEB-RECEIVE refusal). Assumption unchanged: the 16 bytes are opaque (X47, ASSUMED).Cross-repo
None. Overlaps with #5898, #5900, #5975 on
equivalence_cics.py: origin/main merged, small edits.Gates
New + related tests (965 passed),
pr_gates --fast7/7,--ratchets9/9,audit_checkclear, plainpytestfrom/tmpOK.det_port.py check: the forge change moves generated entity text in 105 ports; all re-proved, 104 PROVED.cbsa-crecustDIFFERS: its oracle refusesCEEDAYS: picture not modelled: a non-blank tail(ceedays.cuntouched here), unrelated to this PR. No golden bless needed (no parser change).Sibling gaps
DSNTIAR / CEEHDLR / CEEHDLU refused by different mechanisms; ZECS002 / ZECSNC are assembler; other
text/*media types hit the second-WEB-RECEIVE refusal. More than 5 hand-written files (two engine fixes).Remaining
Slice 3:
cics_census.py surveyshowing ZECS001 with 0 holes, L1 evidence refresh,tests/fixtures/evidence_report/zecs/report.*.det_sweep_coverage/zecs-zecs001.jsonnot added (scheduled job).Lessons for agents
fromRecordchange moves ~105 det-port outputs; rundet_port.py checkearly.WRITE LENGTH(n)with spaces, not zeros.&ID;-style templates need install-symbol substitution on the harness side.🤖 Generated with Claude Code