Proposals / #1048

Recover the signed decimal result consumed as a negative exponent

applied · by Commit-Will · submitted 2026-11-02, applied 2026-11-02

Ghidra v1841 · 3cfe190

Win00413ac0 records uint32_t, but constructs a minus-prefixed decimal result with full NEG EAX. Both actual callers feed decimal exponent suffixes, leave '-' intact, test the exact full result with signed JGE, form its magnitude and select the negative division path. A demonstrated raw0xfffffffe means -2 in these complete producer/consumer routes, not unsigned4294967294. Recover int32_t SStrToInt(char * szText), preserving the selected name, existing char* input, oneStack4, stdcall, noVarargs and locals. This is a signed logical-domain correction supported by negative values, not a normalized-range or same-width alias preference. All1219codeB/482instructions plus actual10.0 constant8B match originalPE/freshDB/separateassembly. No original typedef uniqueness, overflow/parsing-standard, floating rounding, opaque helper effect, runtime/build or backend-application guarantee; refusal holds the whole claim, no fallback.

Changes 1, applied together

1. Set prototype win 0x00413ac0

   signatureuint32_t SStrToInt(char * szText)→int32_t SStrToInt(char * szText)
   convention__stdcall