oh, apparently it's not possible to portably check for string-to-float conversion errors in standard c

A deep dive into the C standard's specification of string-to-float conversion functions (strtod, strtof, strtold, etc.) reveals that portably checking for overflow, underflow, and invalid-input errors is far trickier than the common errno-check idiom suggests. The standard C spec ties error reporting to the math_errhandling macro (errno vs. floating-point exceptions), while POSIX unconditionally requires errno to be set on overflow/underflow, creating a conflict that may make POSIX's own behavior non-conformant to standard C. For invalid input, POSIX only says errno 'may' be set to EINVAL, and glibc and musl actually diverge: glibc never sets errno to EINVAL while musl does, a difference undocumented in the Linux man pages. The piece closes with a concrete checklist of standards-correct ways to detect overflow, underflow, and invalid input across implementations.

Nguồn: https://sebsite.pw/w/20261009-strtod.html. 8 Sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.