|  | The gdtoa code here is based on David M. Gay's original | 
|  | gdtoa source at http://www.netlib.org/fp/ from Jan. 6, | 
|  | 2023. The major changes between the original source and | 
|  | the mingw port here include: | 
|  |  | 
|  | * IBM, CRAY and VAX code removed. | 
|  | * KR_headers, ANSI, Void and Char ifdefs are removed. | 
|  | * gdtoa symbols are prepended with "__". | 
|  | * g_xfmt() uses __fpclassifyl() instead of interpreting | 
|  | the flags bit-wise. | 
|  | * lo0bits() and hi0bits() of misc.c replaced by wrappers | 
|  | to gcc's __builtin_clz() | 
|  | * The double/ulong union renamed from U to dbl_union | 
|  | (grep'ped better..) | 
|  | * A few compiler warning fixes here and there. | 
|  | * A few other insignificant changes (if any..) | 
|  | * set_max_gdtoa_threads functionality removed | 
|  |  | 
|  | MinGW specific compile-time definitions are at the top of | 
|  | gdtoaimp.h and gdtoa.h headers. | 
|  |  |