Skip to content

bundle: five reviewed fixes from the outside review (#746, #747, #748, #749, #750) - #751

Open
MendixMau wants to merge 18 commits into
masterfrom
claude/loving-gates-hauc1n-w1-bundle
Open

MendixMau wants to merge 18 commits into
masterfrom
claude/loving-gates-hauc1n-w1-bundle

Conversation

@MendixMau

Copy link
Copy Markdown
Owner

Direct lane (changes to skills/, bug-logs/, bin/, project-bin/, project-tests/, pipelines)

What changed and why (one paragraph): This branch is master plus the five already-reviewed pull requests merged in, unchanged, so they can be tested and released together instead of one after the other (each adds a line to CHANGELOG.md, so merging them one by one means five rounds of refresh and CI). Nothing here is new: every change is described, with its evidence, in its own pull request. The only hand resolution is CHANGELOG.md, where all five lines are kept under ## Unreleased.

PR Change Independent review
#746 The cutover checkpoint names the unfiled-bugs line and the command to run MERGE at f812af08
#747 check-design-reaches-app.sh normalises CSS and matches whole classes (#670). Can turn a wrongly passing project red. MERGE at 9ed2425f
#748 exec.sh prints one MXTK-RESULT line on every path that reaches a verdict; additive only MERGE at 29ca2b78
#749 Placeholders that span lines are counted; an unfilled multi-line slot now fails build-ready (maintainer decision 2026-11-13) MERGE at 02c2b3e6
#750 The local lane gets explicit database choices, names a missing JDK, snapshots a split model; default lane unchanged MERGE at 6d82cd09

Impact

Row Answer
Process As in the five PRs: two checks become stricter on purpose (#747, #749); one new block when no JDK is installed (#750).
Dependencies The five PRs touch separate files; only CHANGELOG.md overlaps. #748 and #749 both touch scripts that later work will build on (exec.sh, gate-check.sh).
Context cost none: baseline 66,025 → 66,025 of 80,000 (bin/render-routing.sh --check on this branch).
Behaviour n/a beyond the five PRs.
Runtime As measured in the five PRs: +0 s per exec (#748), +0.1 s per gate run (#749), +4 s per design check (#747), one model snapshot per local start of a split model (#750).

Field evidence — see each PR. A composition check of four of the five on a throwaway merge passed every scoped fixture on 2026-11-12; a composition check of all five on this exact branch is running and its result will be posted here before merge.

  • No client data anywhere in the diff
  • Size cap: exceeded by design (20 files); this is five reviewed PRs stacked, each within the cap
  • Test tier reached: per PR; plus scoped fixtures on the combination
  • For new/changed instruments: per PR
  • CHANGELOG.md: five lines under ## Unreleased, one per PR
  • New bug entries: n/a

Generated by Claude Code

The slot count joins lines before matching, so a placeholder wrapped onto a second line is counted. Body excludes '{' so the quoted '{{' in stub line 11 cannot swallow later slots. build-ready now FAILs on a wrapped unfilled slot (was WARN). sync-project is_pure_stub left per-line deliberately.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…matching (#670)

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…the real command

Review fix: the runbook sentence repeated a rule already in an always-loaded file and was attached to the ledger sentence; reverted. The checkpoint now quotes the command gate-check prints and when it prints it. No always-loaded words added.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…m the committed fixture pair

Review fix: the line had landed inside the release-cycle paragraph, where cut-release.sh would not find it.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…stinct lint_new words, more cases

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…ver attempted one

Review fix: on a partial run, and when mxbuild crashed with no baseline (write kept on purpose), the new closing line told the reader to restore the model, contradicting the text above it. It now says no rollback was attempted. The MXTK-RESULT line is unchanged.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…ver scan now FAILs too

Maintainer decision 2026-11-13: an unfilled multi-line placeholder fails build-ready
(supersedes the #509 warning). The perl scan stays only for a slot containing a single
brace, which the joined-lines count cannot match.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
… database yet, and protects a split model

RUN_LOCAL_DB (auto|postgresql|ensure|hsqldb), JDK/Postgres probe with exit 4, snapshot-mpr.sh before a split-model start.
Field run: copy of a real split-model project, no database: master fails, this serves 200 in 46 s.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…e failure line

Review fix: a wrapped placeholder carried its whole hint text (about 300 characters) into the build-ready failure message. Slots over 80 characters now print by name; shorter ones are unchanged.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
… tolerant; scrub fixture path

RUN_LOCAL_DB=auto passes no database option and does not probe TCP; on a
'database not reachable' failure the script names hsqldb / ensure and stack.conf.
hsqldb is explicit only; non-default mode is warned and published as LOCAL_DB.
JDK probe uses find_java_exe, reads the first digit run, warns when unreadable.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…K warns and continues; per-mode db hint

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jQvaonym9V1wKKfAEPmRr
…sign670' into claude/loving-gates-hauc1n-w1-bundle

# Conflicts:
#	CHANGELOG.md
…sult-line' into claude/loving-gates-hauc1n-w1-bundle

# Conflicts:
#	CHANGELOG.md
…aceholders' into claude/loving-gates-hauc1n-w1-bundle

# Conflicts:
#	CHANGELOG.md
…cal-lane' into claude/loving-gates-hauc1n-w1-bundle

# Conflicts:
#	CHANGELOG.md
@MendixMau

Copy link
Copy Markdown
Owner Author

Composition check on this exact branch (765953bf): MERGE. Run by an agent that wrote none of the five changes.

  • The bundle is master plus the five branches and nothing else: for each branch, its non-changelog files are byte-identical here; 20 changed files, each traced to one branch; only CHANGELOG.md is touched by more than one, with exactly five added lines under ## Unreleased, no duplicates, no conflict markers.
  • No interaction found: _common.sh and snapshot-mpr.sh are unchanged by any branch; exec.sh, test-stack-up.sh, gate-check.sh and check-design-reaches-app.sh are each edited by one branch only.
  • Scoped checks on the combination: check-scripts 134/134 and 30/30; render-routing in sync, baseline 66,025 (same as master); check-no-private-citations clean; check-portability clean; test-exec-result-line 72/0; test-exec-gate-retry 52/0; test-exec-lint-diff 9/0; test-exec-crash-rollback 13/0; test-check-stubs-filled pass; test-bug03-gates 99 ok; test-design-reaches-app 70/0; test-stack-up-local 92/0.

Not run: the full suites (CI runs them), macOS, Git Bash.


Generated by Claude Code

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.

2 participants