equivalence: CALL USING items and COMMAREA trailing FILLER compared whole; BMS OCCURS refused (Fixes #4778) by squid-protocol · Pull Request #4780 · squid-protocol/gitgalaxy

squid-protocol · 2026-10-09T10:39:41Z

Fixes #4778. (Was stacked on #4772; that has merged and origin/main is merged in, so the diff is now only #4778.)

What changed

1. CALL USING / LINKAGE compare (tests/tools/equivalence_call.compare_dto → compare_item)

A group USING item is now read by its whole layout (layout_fields(..., occurrences=True)). Every field and every occurrence is compared by name and subscript (PARM-AMT(3)).
Ports that take the item's bytes. A det port now has withCallAreas(byte[]...) (det/program.py). The harness passes each item's bytes before handleCall, BY REFERENCE, and the port gives back what it left there. These are compared like a record (common.diff_records): every field by value and then by bytes, each FILLER as bytes (FILLER@<offset>), and any byte no field covers. When bytes are passed they are what the program runs over; the DTO is still filled on the way out. withCallAreas is added to proof_reach.PROOF_ROOTS, because the harness calls it.
DTO-only ports (model ports). Every named field and every active occurrence is compared. A field the DTO does not map counts as absent: it is compared, never assumed equal. A DTO cannot carry unnamed bytes, so the report lists them under not_compared with that reason.

2. Trailing FILLER

det.cics.Dto.size now covers the whole declared record whenever the record is longer than the DTO's properties reach. This is no longer limited to OCCURS DTOs. It affects CBSA CustctrlDfhcommarea (42→259 bytes) and AcctctrlDfhcommarea (46→98), and GenApp Lgicvs01 / Lgsetup / Lgwebst5 (24→84) and Lgipvs01 (32→90). With the full size, a LINK passes those bytes, a RETURN or XCTL hands them on (DetCics.commareaOut), and EIBCALEN matches the record length.
COMMAREA compare: decode_record(..., unnamed=True) reads each FILLER as hex. FILLER is compared when both sides give bytes: the COBOL side always does, and the Java side does for a det port LINKed with the area or a COMMAREA longer than its DTO. When the Java side gives back a DTO, FILLER is declared not_compared with the reason. An all-LOW-VALUES FILLER in a task that has no COMMAREA is undefined storage (X12) and is masked, like the named fields.

3. BMS OCCURS: screen_layout refuses a symbolic map with OCCURS (Unsupported: BMS OCCURS), because its screen compare would read only the first occurrence. test_no_case_map_has_occurs_today checks every case's maps, so the first map with OCCURS fails that test.

Mutation proof (the weakness, shown)

The mutant is the det port of mortgage-nbrvl, changed in two ways:

The contract DTO no longer maps EPSPARM-RETURN-ERROR.
The service writes 'X' into that field's last byte after the program runs.

Results:

origin/main harness: EPSNBRVL CALLS: 15/15 calls equal, proven.
this branch: 0/15 calls equal, EPS-NUMBER-VALIDATION.EPSPARM-RETURN-ERROR: cobol ' …' java ' …X', caught.

Re-sweeps (on the merged tree)

proof_sweep.py --det-only (Db2 local): 77/80 proven. The 3 that are not proven are the baseline's (carddemo-cardupdate, carddemo-intcalc-generated, mortgage-mpmt). Result: "sweep: as expected", no newly failing case. FILLER is now compared in cbsa-custctrl, cbsa-dbcrfun and carddemo-cardupdate.
proof_sweep.py --model-only: 28/28 proven, "as expected".
Coverage ledger re-measured (det_coverage_ledger.py update): the numbers did not change. Only fingerprints were refreshed: harness, oracle, and cbsa-inqacccu's case.
evidence.py status --ci: rc 0, so no record is blocking-stale. A harness change makes records stale only through the scheduled refresh.

Gates

pr_gates.py --fast passes 5/5 and --ratchets passes 7/7 (JDK 17).

🤖 Generated with Claude Code

https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

… table travels (Fixes #4765) The harness's record layout listed an OCCURS group's fields once (the first occurrence), so a COMMAREA's table was compared at occurrence 1 only: a cbsa-inqacccu port that never wrote account 3's type was PROVEN on main. - layout_fields(..., occurrences=True): every occurrence a field, by name and subscript (COMM-ACC-TYPE(3), nested CELL(2,4)), with its OCCURS DEPENDING ON constraints; a COMMAREA is read by it, each side by its own ODO count (active_fields). A file's record names each occurrence (they were compared only as "bytes outside the layout"). - a scenario's COMMAREA is the whole record (it was cut at the first occurrence's end: INQACCCU's EIBCALEN was 123 of 1985 bytes, accounts 2..N written past it), every occurrence INITIALIZEd. - a LINKed task gets its COMMAREA's bytes too (CicsTask.withLinkArea); a port that takes them (a det port) is read from them; a DTO's values are the first occurrence (first_occurrences), the rest absent -- compared. - det: a DTO whose OCCURS fields appear once spans its declared record with or without a POINTER (LINK span, EIBCALEN), and RETURN / XCTL pass its bytes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

…4765) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

# Conflicts: # tests/equivalence/det_sweep_coverage.json

…hole; BMS OCCURS refused (#4778) - compare_dto: a group USING item is read by its whole layout, every occurrence by subscript; a port taking the items' bytes (det: withCallAreas) gives them back and they are compared as bytes (each FILLER too); a DTO-only port has every named field compared and its unnamed bytes listed as not compared, with the reason. - det Dto.size spans the declared record when it ends in a FILLER (CBSA CUSTCTRL, GenApp LGICVS01/LGIPVS01/...); COMMAREA FILLER compared as bytes when both sides are bytes, declared not compared against a DTO. - screen_layout refuses a symbolic map with OCCURS by name. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

…LLER compare (#4778) Numbers unchanged; harness fingerprints and measured_at refreshed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

# Conflicts: # gitgalaxy/tools/cobol_to_java/det/cics.py # tests/equivalence/det_sweep_coverage.json # tests/tools/equivalence_cics.py

github-actions · 2026-10-09T11:43:06Z

🐦‍⬛ Muninn Security Scan

✅ No security issues found.

🐦‍⬛ Powered by Muninn · Skald Lab

Full det sweep on the merged tree: 77 proven, the 3 baseline cases not; numbers unchanged, harness / oracle fingerprints refreshed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

# Conflicts: # tests/equivalence/det_sweep_coverage.json

…-swept next) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…n, as expected; coverage unchanged) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…its full det sweep Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

squid-protocol · 2026-10-09T21:43:25Z

CI shepherd: this PR conflicts with main, so no checks ran. Merge origin/main into it and regenerate generated files with their tools (never hand-merge them; see .claude/agents/gitgalaxy-pr-worker.md), then push.

squid-protocol · 2026-10-09T22:15:22Z

CI shepherd: this PR conflicts with main, so no checks ran. Merge origin/main into it and regenerate generated files with their tools (never hand-merge them; see .claude/agents/gitgalaxy-pr-worker.md), then push.

… split --base + update from the sweep Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

squid-protocol · 2026-10-09T23:28:32Z

CI shepherd: this PR conflicts with main, so no checks ran. Merge origin/main into it and regenerate generated files with their tools (never hand-merge them; see .claude/agents/gitgalaxy-pr-worker.md), then push.

…e; ledger re-applied from its sweeps Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

squid-protocol and others added 5 commits October 9, 2026 04:56

det sweep coverage ledger: re-measured under the whole-table compare (#…

78c33cc

…4765) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MQNoe4wPJgr7dbUvs3DG6D

Merge remote-tracking branch 'origin/main' into fix/4765-occurs-compare

64ea762

# Conflicts: # tests/equivalence/det_sweep_coverage.json

Merge remote-tracking branch 'origin/main' into fix/4765-occurs-compare

214475a

squid-protocol added bug Unintended behavior or logic failure in the engine legacy-modernization COBOL refractor, dead-code extraction, and JCL forging testing Unit, integration, and E2E pipeline verification labels Oct 9, 2026

squid-protocol added this to the Training Montage 🏋️ — Prepping to one-shot our 4th CICS to Java translation attempt milestone Oct 9, 2026

squid-protocol and others added 2 commits October 9, 2026 07:29

Merge remote-tracking branch 'origin/main' into fix/4778-call-compare

f05487a

# Conflicts: # gitgalaxy/tools/cobol_to_java/det/cics.py # tests/equivalence/det_sweep_coverage.json # tests/tools/equivalence_cics.py

squid-protocol changed the title ~~WIP: equivalence: CALL USING items and COMMAREA trailing FILLER compared whole (Fixes #4778)~~ equivalence: CALL USING items and COMMAREA trailing FILLER compared whole; BMS OCCURS refused (Fixes #4778) Oct 9, 2026

squid-protocol and others added 5 commits October 9, 2026 08:44

Merge remote-tracking branch 'origin/main' into fix/4778-call-compare

7717416

# Conflicts: # tests/equivalence/det_sweep_coverage.json

Merge remote-tracking branch 'origin/main' into fix/4778-call-compare

b2d2168

# Conflicts: # tests/equivalence/det_sweep_coverage.json

Merge remote-tracking branch 'origin/main' into fix/4778-call-compare

316d3ae

# Conflicts: # tests/equivalence/det_sweep_coverage.json

Merge remote-tracking branch 'origin/main' into fix/4778-call-compare

c2a3506

# Conflicts: # tests/equivalence/det_sweep_coverage.json

Merge origin/main into fix/4778-call-compare; main's ledger taken (re…

70c272b

…-swept next) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

squid-protocol mentioned this pull request Oct 9, 2026

equivalence: compare BMS map OCCURS fields by occurrence -- CBSA BNK1CCA's BNK1ACC (ACCOUNTD OCCURS 10) cannot be cased once #4780 refuses OCCURS maps #4807

Closed

det sweep ledger: full det re-sweep after the main merge (77/80 prove…

d497680

…n, as expected; coverage unchanged) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Merge origin/main into fix/4778-call-compare; ledger re-applied from …

7fcb6f4

…its full det sweep Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

Merge origin/main (#4811 per-case ledger) into fix/4778-call-compare;…

d5a27ac

… split --base + update from the sweep Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

Merge origin/main (#4810 CBSA cases, #4826) into fix/4778-call-compar…

2fa5c63

…e; ledger re-applied from its sweeps Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TBJdDBRtJis5hzE244NQKV

squid-protocol marked this pull request as ready for review October 9, 2026 23:59

squid-protocol merged commit e60b93f into main Oct 9, 2026
51 checks passed

squid-protocol deleted the fix/4778-call-compare branch October 9, 2026 23:59

squid-protocol mentioned this pull request Oct 10, 2026

det: a LINK gives the caller's COMMAREA bytes back as the callee left them (Fixes #4806) #4818

Merged

squid-protocol mentioned this pull request Oct 18, 2026

Siblings of #5076: CICS programs whose equivalence case is refused by the harness/facade #5089

Open

4 of 7 tasks

Navigation Menu

Repository navigation

equivalence: CALL USING items and COMMAREA trailing FILLER compared whole; BMS OCCURS refused (Fixes #4778) - #4780

equivalence: CALL USING items and COMMAREA trailing FILLER compared whole; BMS OCCURS refused (Fixes #4778)#4780
squid-protocol merged 17 commits into
mainfrom
fix/4778-call-compare

squid-protocol commented Oct 9, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 9, 2026

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

squid-protocol commented Oct 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What changed

Mutation proof (the weakness, shown)

Re-sweeps (on the merged tree)

Gates

Uh oh!

github-actions Bot commented Oct 9, 2026

🐦‍⬛ Muninn Security Scan

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

squid-protocol commented Oct 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

squid-protocol commented Oct 9, 2026 •

edited

Loading