zp-large-struct-by-value — inconclusive
Status
inconclusive — no evidence/zp-large-struct-by-value.jsonl. Do not invent one.
Reason
Large-struct-by-value vs pointer microbench under Zig 0.17.0 -Ofast / ReleaseFast was elided (~tens of ns total). Numbers do not represent the smell; citing them would be dishonest.
Why no JSONL
Optimizer removed the work; ratio is noise.
Unblock
A harness that forces observable copies (opaque / volatile / black-box sink, or a real call boundary the optimizer cannot erase) under the optimize mode we care about. Until then the rule stays off.