hyf

commit e323906b602c5ac878e4136b4cb35a271a8effc3
parent b065cef0baf1fdb1dd91c6156e9d62dcdc4b5c39
Author: triesap <tyson@radroots.org>
Date:   Wed, 23 Sep 2026 03:48:52 +0000

test(hyf): prove provider report-reader overflow rejection (C002D review 4)

Fourth independent read-only review of b065cef passed with one non-blocking
INFO: the shared report reader's overflow cap had no dedicated provider-path
control. Add an oversize report control (line content beyond
STRICT_MAX_REPORT_BYTES) and prove both provider reap paths reject it with
ready_output_overflow. Test-only; no criterion or product change.

Re-verified on final source: provider-helpers 80/80, repo-local-process 9/9,
test-stdio 56/32/24/0 with identical D16 names and reasons, format/architecture/
build green.

Diffstat:
M
tests/test_provider_helpers.mojo
 | 
9
+++++++++

1 file changed, 9 insertions(+), 0 deletions(-)
diff --git a/tests/test_provider_helpers.mojo b/tests/test_provider_helpers.mojo
@@ -44,6 +44,7 @@ from parent_lifecycle import (
     write_raw_bytes,
 )
 from strict_fixture import (
+    STRICT_MAX_REPORT_BYTES,
     ConnectionReader,
     ExchangeScript,
     FramedRequest,
@@ -1675,6 +1676,14 @@ def test_report_stream_controls_both_providers() raises:
         0,
         "duplicate_field",
     )
+    var oversize = "result ok phase=complete case="
+    while (
+        oversize.byte_length() + tail.byte_length()
+        <= STRICT_MAX_REPORT_BYTES + 1
+    ):
+        oversize += "z"
+    oversize += tail
+    _report_controls(oversize, 0, "ready_output_overflow")
 
 
 def test_startup_failure_cleanup_ownership_is_truthful() raises: