Repository navigation
Det ports declare themselves; the aperture admits declared ports past the generated-noise gates - #4164
Merged
Conversation
…t the generated-noise gates The det emitter writes a first-line comment on each service it emits (// gitgalaxy-det-port: COBOL <PROGRAM> (<source>), ...) and puts storage-image literals one 400-character piece per line. GitGalaxy's aperture reads that line (declared_port: .java, first line only) and lifts Gates 4.1 (to DECLARED_PORT_MAX_LINE_LENGTH, 5000), 4.3 and 5.1 for it. Binary, payload, amalgamation, size and secrets gates are untouched: a forged header can only admit code to the scan, never hide it. Before: a default scan excluded 45 of 49 det ports; after: 49 of 49 scanned. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ
Contributor
squid-protocol
added a commit
that referenced
this pull request
Oct 2, 2026
…4163/#4164 Spearman now gives tied values their average rank (I/O rho 0.24, not -0.03); records the I/O and IPC agreement after #4163 (0.87 / 0.84). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ
squid-protocol
added a commit
that referenced
this pull request
Oct 2, 2026
* docs(wiki): COBOL <-> Java scan parity -- what GitGalaxy sees before and after a port 49 programs from CardDemo, CBSA and GenApp scanned as COBOL, as the proven deterministic Java port and (23 CardDemo programs) as the model-written port: structure keeps its rank (functions rho 0.94, branches 0.83, mutations 0.76), size and token mass do not, I/O moves into the cobolrt runtime, risk scores rank within one language only. Method, tables, scanner and translation lessons, and the follow-ups in progress. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ * docs(wiki): scan parity -- tie-corrected Spearman, follow-ups #4162/#4163/#4164 Spearman now gives tied values their average rank (I/O rho 0.24, not -0.03); records the I/O and IPC agreement after #4163 (0.87 / 0.84). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ * docs(wiki): scan parity -- COTRTLIC keeps one hole (dynamic CALL); 48 of 50 cases prove Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Owner
Author
|
Full det sweep with the header and wrapped images ( 🤖 Generated with Claude Code |
This was referenced Oct 2, 2026
squid-protocol
added a commit
that referenced
this pull request
Oct 2, 2026
…Db2, oracle register, 05-19) (#4166) * docs: COBOL-to-Java capability docs brought up to date (48 det ports proven, Db2, oracle register) - README, det_port_design, porting loop, cobol_to_java README: 48 programs proven as det ports (27 CardDemo, 8 CBSA, 13 GenApp; 17 cases on Db2), the two cases not proven on purpose, TRUNC(STD), the Db2 unit of work and its limits, proof_sweep.py, the oracle register's C9 / D1 / M2 / Q3. - statement_inventory.md regenerated (2026-10-02): ASSIGN PROGRAM / INVOKINGPROG, GET COUNTER, DELAY, ENQ / DEQ left the holes; EXEC SQL is counted H by definition though det-port translates it. - oracle_assumptions.md: Q3 (one Db2 unit of work per CICS task, batch autocommits) and Q4 (positioned UPDATE / DELETE no longer refused) corrected. - New wiki page 05-19 (proven ports, harness, register, limits); cookbooks and 05-02 / 05-04 / 05-05 / 05-06 / 01-04 / map-cobol-monoliths say how logic is ported and proven, and drop stale claims (PostgreSQL-only, LLM-only logic, TODO stub text). - cobol-modernization skill: closed ownership table replaced; points to the det-port skill. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ * docs(wiki): scan parity -- #4164 merged: a plain scan admits all 49 det ports Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ * docs(wiki): 05-19 links scan parity and the declared-port header (#4164) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ * docs: det_survey.md regenerated (2026-10-02): 97.8% of 17,224 statements, no EXEC SQL holes Re-run of tests/tools/det_survey.py over the six corpora (sequential; javac plugin jars from another project's ~/.m2 left off the classpath). 21 programs proven outside CardDemo. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
A default GitGalaxy scan excluded 45 of the 49 det ports as machine-generated noise: 4 via the saturation gate (lines over 500 characters) and the rest via the lexical-monotony gate. A migrated estate would look almost empty.
What
det/program.py): each emitted service now has a first-line comment,// gitgalaxy-det-port: COBOL <PROGRAM> (<source>), translated by rule, statement for statement. It also writes storage-image Base64 literals as one 400-character piece per line instead of 30,000-character runs. Both are formatting only.core/aperture.py):declared_port()reads that line, only from the first line of a.javafile. A declared port passes three gates:DECLARED_PORT_MAX_LINE_LENGTH(5000);Validation
tests/core_engine/test_aperture.py: 24 passed.det_port.py run carddemo-report cbsa-inqacc genapp-lgicdb01all PROVED with the header and wrapped images. Corpt00c had a 106K-character image line.galaxyscope(no wrapper) on all 49 freshly emitted det ports: CardDemo 28/28, CBSA 8/8, GenApp 13/13 scanned, 0 excluded.proof_sweep.py --det-only): result posted below.tests/ruff_audit.py --ci: no new findings.🤖 Generated with Claude Code
https://claude.ai/code/session_01RZE6us7fUL6Hynk59rWLvZ