Skip to content

jcl: re-baseline for gitgalaxy#2610 — args 13, live comment surface, 13R/7A → 6R/8A - #4

Merged
squid-protocol merged 2 commits into
mainfrom
feat/jcl-2610-rebaseline
Aug 31, 2026
Merged

jcl: re-baseline for gitgalaxy#2610 — args 13, live comment surface, 13R/7A → 6R/8A#4
squid-protocol merged 2 commits into
mainfrom
feat/jcl-2610-rebaseline

Conversation

@squid-protocol

Copy link
Copy Markdown
Owner

Companion to squid-protocol/gitgalaxy#2611 (closes gitgalaxy#2610), for gitgalaxy#2581 / epic #2560.

Draft until gitgalaxy PR #2611 merges — this repo's CI checks out gitgalaxy main, so the jcl gate here fails until the engine fix lands there. Everything verified locally against the #2611 branch: strict gate PASS (76 assertions), every plant lands exactly in --report mode.

What moved

metric before after
args 1 🔴 13 (median)
safety / safety_bypasses / telemetry 0 🔴 2 / 2 / 2 (median)
fragile_debt / planned_debt 0 🔴 1 / 1 (median)
comment_lines 5 🔴 15 🟡 (real //* mass now measured)
jcl overall 13🔴 / 7🟡 6🔴 / 8🟡, avg consistency 75%→76%

Remaining reds are ledgered intended morphology (cleanup/doc/test/globals — see jcl-2610-rebaseline-residual-morphology) or cross-cutting median inflation (#2545/#2546), plus downstream §3 risk metrics.

🤖 Generated with Claude Code

https://claude.ai/code/session_01PUfXMUECX4Vq1vqh9mMQqt

squid-protocol and others added 2 commits August 31, 2026 14:22
…MSG*/debt rules) + args to 13 via PARM=

Shell re-authoring (spec-compliant, planted intent unchanged in kind):
- every EXEC step now carries PARM= -> args 4/3/3/3 = 13 (the spec's 'one
  argument per function where the language expresses arguments' -- JCL
  always could, the original authoring under-used it)
- plants for the new #2610 rules: safety 2 (COND=(8,LE)/(4,LT) on the
  a.jcl steps), safety_bypasses 2 (COND=EVEN/ONLY in b.jcl), telemetry 2
  (MSGLEVEL=/MSGCLASS= as EXEC-proc symbolic overrides on PROBETEL),
  fragile/planned debt 1/1 (//* HACK: / //* TODO comments in c.jcl)
- verify_language.py jcl --report: every plant lands exactly; strict gate
  PASS 76 assertions

Ledger: new validated entry jcl-2610-rebaseline-residual-morphology
covering the intended-zero residuals (cleanup io-overlap decision, doc/
test/globals absence, DD DSN= dependency capture, comment_lines blank-
line proxy, branch/state_mutation median inflation by #2545/#2546).

Bias artifacts regenerated (46 languages, engine at gitgalaxy PR #2611):
jcl 13R/7A -> 6R/8A; avg consistency 75% -> 76%.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PUfXMUECX4Vq1vqh9mMQqt
@squid-protocol

Copy link
Copy Markdown
Owner Author

The failing verify-languages check is the expected engine dependency, not a regression: the 5 mismatches (safety 2→0, safety_bypasses 2→0, telemetry 2→0, fragile_debt/planned_debt 1→0) are exactly the five signals whose jcl rules only exist in squid-protocol/gitgalaxy#2611 — this workflow checks out gitgalaxy main, which doesn't have it yet. Everything the old engine can measure passes, including the args 13 re-plant. Re-run after #2611 merges and this goes green.

@squid-protocol
squid-protocol marked this pull request as ready for review August 31, 2026 19:28
@squid-protocol
squid-protocol merged commit 7dd8e17 into main Aug 31, 2026
1 of 2 checks passed
squid-protocol added a commit that referenced this pull request Aug 31, 2026
…solution

Conflict resolution per the ledger playbook: took main's deviation_ledger.json
(36 entries, incl. PR #4's jcl-2610-rebaseline entry) and re-applied the
cobol-area-b-bare-token-paragraph still_reproduces flip programmatically --
never hand-merged the JSON.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PUfXMUECX4Vq1vqh9mMQqt
@squid-protocol
squid-protocol deleted the feat/jcl-2610-rebaseline branch August 31, 2026 21:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant