[kddejong]

Fixes #4877. Supersedes #4912, which fixed this by making E6004 fire; this removes it instead.

Problem

E6004 (src/cfnlint/rules/outputs/Pattern.py, "Outputs have appropriate names") has no match and no validate method — it is a docstring and five class attributes. It never fires.

rules/outputs/Configuration.py maps "pattern": "E6004" in its rule_set, but that mapping keys off err.validator, and the Outputs name rule is expressed as patternProperties plus additionalProperties: false in data/schemas/other/outputs/configuration.json. The pattern keyword appears nowhere in that schema, so err.validator == "pattern" can never occur there. A bad output name fails on additionalProperties and is reported by E6001, and --ignore-checks E6004 does nothing.

Measured on upstream/main:

$ cfn-lint -f parseable -t bad_name.yaml
bad_name.yaml:5:3:5:12:E6001:'My_Output' does not match any of the regexes: '^[a-zA-Z0-9]+$'

$ cfn-lint -f parseable -i E6004 -t bad_name.yaml     # no-op
bad_name.yaml:5:3:5:12:E6001:'My_Output' does not match any of the regexes: '^[a-zA-Z0-9]+$'

$ cfn-lint -f parseable -i E6001 -t bad_name.yaml     # silent — the finding really is E6001

Why remove rather than wire it up

The discrimination E6004 implies — separating a bad logical ID from a nested unknown property — exists nowhere else in the codebase. Measured on upstream/main:

Section	Bad logical ID	Nested unknown property
Resources	E3001	E3001
Parameters	E2001	E2001
Mappings	E7001	—
Outputs	E6001	E6001

No section separates the two; each reports its own section rule. Outputs is also the only section whose rule_set maps pattern to a name rule at all — Resources and Mappings have no pattern entry, and Parameters maps pattern to E2003, which is AllowedPattern validation and genuinely fires.

So E6004 assumes a design this codebase does not have. Validation is already correct; only the dead rule is wrong.

Its docstring is also wrong — "Check if Outputs strings are less than the max length" on a rule named "Outputs have appropriate names" — which is further evidence nobody has touched it.

The change

1. Delete src/cfnlint/rules/outputs/Pattern.py.
2. Remove "pattern": "E6004" from self.rule_set in src/cfnlint/rules/outputs/Configuration.py. Every other entry is untouched — in particular "propertyNames": "E6011" and "maxLength": "E6011", which carry the working 255-character length check.
3. docs/rules.md: drop the E6004 row and decrement the headline count 334 → 333 (the generator computes it as len(sorted_rules) + 3, maintenance.py:63; running cfn-lint --update-documentation on this branch produces 333).

No test, snapshot, .cfnlintrc or fixture referenced E6004 — it had no test coverage at all. grep -rn E6004 src/ test/ docs/ scripts/ returns zero hits after the change.

Verification

No user-visible validation changes. Before/after on upstream/main vs this branch, identical in every case:

Check	Result (both)
Output named My_Output	E6001 ... does not match any of the regexes: '^[a-zA-Z0-9]+$' at 5:3:5:12
Unknown property in an output	E6001 Additional properties are not allowed ('Bogus' was unexpected) at 7:5:7:10
Unknown property inside Export	E6001, same message, at 12:7:12:12
256-character output name	E6011 ... is longer than 255 at 5:3:5:259
Outputs as a list	E6003 [{'Value': 'foo'}] is not of type 'object'
Output missing Value	E6002 'Value' is a required property
201 outputs	E6010 'Outputs' has more than 200 properties
Valid template	silent, exit 0

--list-rules no longer lists E6004 and the diff against main is exactly that one rule's two lines (it prints id+shortdesc then description, 662 → 660). -i E6004 now behaves like any unrecognised rule ID: the CLI accepts it silently with no warning or error — confirmed identical to -i E9999999 on main.

Good-fixture sweep, run by hand because test/integration/test_good_templates.py is a curated ~20-template list and does not sweep the directory:

cfn-lint -f parseable -t "test/fixtures/templates/good/**/*.yaml"   # 98 templates, 162 findings
cfn-lint -f parseable -t "test/fixtures/templates/good/**/*.json"   # 9 findings

Noise floor established first by running main against itself three times: byte-identical, 0 differing lines (E3019/E3043, which render Python sets into their messages, do not appear in these fixtures). Against that floor, main vs this branch is byte-identical for both yaml and json, and an order-insensitive check (rule-ID plus file:line multiset) also matches.

Full suite, compared as failure sets rather than counts:

main:   6 failed, 3346 passed, 4 deselected
branch: 6 failed, 3346 passed, 4 deselected

Failure sets are identical. All 6 are pre-existing and caused by missing optional dependencies in the local environment (pydot, junit-xml, sarif), not by this change. The pass count is unchanged at 3346, confirming no test was removed.

pre-commit run --all-files passes (ruff, ruff-format, isort, bandit, mypy, AST/JSON/YAML checks).

Backwards compatibility

A user suppressing E6004 via ignore_checks was suppressing nothing, so the suppression can simply be dropped — it will not start reporting anything new. Users who wanted to suppress the output-name check need E6001, which is where the finding has always come from.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.73%. Comparing base (588d3c5) to head (4328fc9).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4983      +/-   ##
==========================================
- Coverage   94.74%   94.73%   -0.01%     
==========================================
  Files         442      441       -1     
  Lines       16145    16138       -7     
  Branches     3138     3138              
==========================================
- Hits        15296    15289       -7     
  Misses        469      469              
  Partials      380      380              

Flag	Coverage Δ
unittests	94.73% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.