Repository navigation
Adobe CF (2023 + 2025): cfheader(attributeCollection) throws 'Failed to add HTML header' — blocks all tests #2741
Description
Describe the bug
On Adobe ColdFusion 2023, every request to the test endpoint (and likely every Wheels request in general) fails with:
Failed to add HTML header.
ColdFusion was unable to add the header you specified to the output stream. This is probably because you have already used a cfflush tag in your template or buffered output is turned off.
The error occurred in /wheels-test-suite/vendor/wheels/Global.cfc: line 123
Called from /wheels-test-suite/vendor/wheels/events/EventMethods.cfc: line 113
Called from /wheels-test-suite/vendor/wheels/Global.cfc: line 248
Called from /wheels-test-suite/vendor/wheels/Global.cfc: line 41
Called from /wheels-test-suite/public/Application.cfc: line 308
121 : cfheader(attributeCollection = "#local.args#");
122 : } else {
123 : cfheader(attributeCollection = "#arguments#");
124 : }
The compat-matrix Adobe 2023 job records 0 pass / 0 fail / 0 err because no test request gets past application boot — every spec file's HTTP request returns the Wheels error HTML page rather than the test results JSON.
Reproduction
- Run
tools/test-matrix.sh adobe2023 mysql(any DB) - Watch every result fail JSON parsing — the body is the Wheels error page
- Or:
curl http://localhost:62023/wheels/core/tests?db=mysql— same error
Evidence
Compat-matrix run 25978222394 (develop, 2026-05-17). Adobe 2023 artifact (test-results-adobe2023) contains HTML for every DB.
Suspected cause
The cfheader(attributeCollection = "#arguments#") call at vendor/wheels/Global.cfc:123 — passing arguments directly. Adobe 2023 likely rejects this shape (maybe expects a struct rather than the arguments scope, or stricter validation than Lucee/BoxLang/Adobe 2021).
The call site is on the dispatch path (EventMethods.cfc:113 → Global.cfc:248 → Global.cfc:41), so every controller action hits it. Lucee 6/7 and BoxLang pass through without issue.
Expected behavior
cfheader(attributeCollection=…) should work cross-engine. Either the call shape needs adjustment for Adobe 2023, or the helper needs to bridge arguments to a plain struct before calling cfheader.
Pending: confirm Adobe 2025 status
The same run's adobe2025 job was still in progress at filing time. If 2025 hits the same regression, this issue covers both versions.
Desktop
- OS: Any
- CFML Engine: Adobe ColdFusion 2023
Activity
wheels-bot commentedon May 17, 2026
Wheels Bot — Triage
Classified as bug.
Suspected layer
controller— see.ai/wheels/controllers/for canonical patterns, and.ai/wheels/cross-engine-compatibility.mdfor the Adobe-CF-specific gotchas that apply here. The defect is in the global$headerhelper that controllers and the dispatch path call to emit HTTP response headers.Fix sketch
vendor/wheels/Global.cfc:112-125— the$header()function callscfheader(attributeCollection = "#arguments#"), passing theargumentsscope directly. Adobe CF 2023 rejects this shape (Lucee 6/7 and BoxLang accept it). The same function already has the pattern needed for the fix: thestatusTextbranch above copiesargumentsinto a plainlocal.argsstruct before callingcfheader. Apply that same copy unconditionally so every call path handscfheadera plain struct, not theargumentsscope. One file, no API change, no caller updates. Verify on Adobe 2023 viatools/test-matrix.sh adobe2023 sqlite(and re-check Adobe 2025 once that job finishes — the issue body flags it as pending).Confidence:
mediumReport and fix location are unambiguous and the shape is mechanical, but cross-engine concern (Adobe-vs-Lucee-vs-BoxLang
cfheaderparsing) fires the auto-downgrade rule — at mostmedium.<!-- wheels-bot:triage:2741 -->
<!-- wheels-bot:triage-class:bug -->
<!-- wheels-bot:triage-confidence:medium -->
wheels-bot commentedon May 17, 2026
Wheels Bot — Fix proposed
Draft PR: #2751
Spec:
vendor/wheels/tests/specs/global/headerSpec.cfc— covers name/value, statuscode, charset, and statuscode+statusText call shapes. Passes locally on Lucee 7 both before and after the fix (Lucee already accepts theargumentsscope), so it documents the cross-engine contract; the failure manifests on Adobe 2023 only and will be verified by the compat-matrix job once the PR runs CI.Implementation:
vendor/wheels/Global.cfc—$header()now copies theargumentsscope into a plain struct (and stripsstatusTextfor Adobe CF 2025 compatibility) before invokingcfheader(attributeCollection=...), collapsing the previous two-branch shape into one uniform code path.A human review is required before merge. Reviewer A and Reviewer B will weigh in shortly. Doc updates (MDX guides,
.ai/wheels/,CLAUDE.md) are handled separately bybot-update-docs.yml.docs: document attributeCollection + arguments scope gotcha on Adobe …
docs: document attributeCollection + arguments scope gotcha on Adobe …
docs: document attributeCollection + arguments scope gotcha on Adobe …
bpamiri commentedon May 18, 2026
Adobe 2025 update — same regression
The Adobe 2025 job from run 25978222394 finished after this issue was filed. Confirmed: Adobe 2025 hits the exact same
cfheader(attributeCollection)failure atvendor/wheels/Global.cfc:123on every DB (cockroachdb, mysql, oracle, postgres, sqlite, sqlserver — all return the Wheels error HTML page).So both Adobe 2023 and Adobe 2025 are wholesale blocked by this regression. Scope of this issue should be read as "Adobe ColdFusion (2023 + 2025)" rather than 2023-only.
[-]Adobe 2023: cfheader(attributeCollection) throws 'Failed to add HTML header' — blocks all tests[/-][+]Adobe CF (2023 + 2025): cfheader(attributeCollection) throws 'Failed to add HTML header' — blocks all tests[/+]fix(dispatch): copy arguments to a plain struct in $header() for Adob…