Rongsong Shen
e48e222bd1
Adds 10 tests in a new test_repl_resource_table.c covering info worker's/info resource's resource-listing table/grouping layout, units_mode-aware rendering (raw/human/both, including swp's MB->bytes conversion and a no-mutation check on the underlying stored value), whole-number-vs-fractional formatting, and info worker's assigned-job listing adopting the real repl_fmt_jobid_range() (Requirements 6.4-6.6, 4.5). All 10 fail as expected against today's flat, unconditional per-resource fprintf loop and uncompressed listing. Found and fixed a genuine ASAN stack-use-after-return in the test's own fixture code (not production): WORKERS_PUSH's handler moves a raw pointer into a job's resvals rather than deep-copying, so the fixture must heap-allocate rather than stack-allocate the structures it feeds through that path - independently reproduced by review. New file rather than extending test_repl_breakpoints_inspection.c: landing this content there first pushed it to 3215 lines, 3.2x this file family's own ~1000-line split trigger already used to justify splitting test_repl_round_rollup.c/test_repl_event_echo.c for structurally identical reasons. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YAWMhfRJFnmmnG8h97yUDd |
||
|---|---|---|
| .. | ||
| settings | ||
| specs | ||
| steering | ||