In NEC-2 semantics RP is an execute request, same as XQ — and most decks in the wild rely on that: NEC-2 manual examples, 4nec2 and EZNEC exports typically contain no XQ at all and end with EX … / RP … / EN.
On current main (b58c4ad), such a deck runs to exit 0 and writes only the comments / structure / segmentation blocks — no matrix fill, no antenna input parameters, no radiation pattern. Nothing in the output or the exit status indicates the solve never happened.
Minimal reproduction (8 lines)
CE dipole, RP as the execute card (no XQ)
GW 1 11 0 -5 10 0 5 10 1e-3
GE 0
FR 0 1 0 0 14 0
EX 0 1 6 0 1 0
RP 0 19 37 1000 0 0 5 5
EN
onec -i deck.nec -o out.txt → exit 0, output ends after SEGMENTATION DATA. No input parameters, no pattern.- nec2c 1.3 on the same deck → full solve: input parameters (64.663 −56.474j) + radiation pattern.
- Add an
XQ before the RP and onec produces both blocks, matching nec2c.
Impact
Found while running the LD-verification corpus for #11: 432 of 492 real-world loaded decks are RP-driven and currently return silently empty results. Arguably higher-impact than #11 — a crash is loud; this is the quiet kind of failure (exit 0, plausible-looking truncated output).
A one-line string check in CI ("output contains ANTENNA INPUT PARAMETERS") on an RP-terminated smoke deck would pin this class — folds naturally into the golden-deck matrix proposed in #15.
In NEC-2 semantics
RPis an execute request, same asXQ— and most decks in the wild rely on that: NEC-2 manual examples, 4nec2 and EZNEC exports typically contain noXQat all and end withEX … / RP … / EN.On current main (b58c4ad), such a deck runs to exit 0 and writes only the comments / structure / segmentation blocks — no matrix fill, no antenna input parameters, no radiation pattern. Nothing in the output or the exit status indicates the solve never happened.
Minimal reproduction (8 lines)
onec -i deck.nec -o out.txt→ exit 0, output ends after SEGMENTATION DATA. No input parameters, no pattern.XQbefore theRPand onec produces both blocks, matching nec2c.Impact
Found while running the LD-verification corpus for #11: 432 of 492 real-world loaded decks are RP-driven and currently return silently empty results. Arguably higher-impact than #11 — a crash is loud; this is the quiet kind of failure (exit 0, plausible-looking truncated output).
A one-line string check in CI ("output contains ANTENNA INPUT PARAMETERS") on an RP-terminated smoke deck would pin this class — folds naturally into the golden-deck matrix proposed in #15.