The definition of vsnprintf as _vsnprintf didn't seem to hurt the build
(as well, for Borland)

Mingw32 is _WIN32 and GCC, and we need some way to get the free()
prototype. :D

Still using BOOL where we need Win32 API compatibility.
Using JSBool instead of BOOL or bool where it matters.
Changed most relevant TRUE/FALSE to true/false too (though it's not as critical).

You shouldn't need to #include <stdbool.h> anywhere now - gen_defs.h should do that automatically/correctly based on the language/version/tool. In C23, stdbool.h isn't even needed for bool/true/false definitions (they're keywords), so we don't bother including stdbool.h in that case.
Microsoft didn't define __STDC_VERSION__ in their older tool chains (even though they were C99 compatible and had stdbool.h), so we use a _MSC_VER check to know that there's a stdbool.h we should use in that case.

For other/old compilers (e.g. Borland C) we #define bool/true/false following the pattern of stdbool.h (doesn't use a typedef).

I didn't convert UIFC yet.

This addresses issue #698

Where ever we assign the realloc() result to the same pointer we pass, use this new function instead to eliminate the cppcheck error reported by Nelgin:
Common realloc mistake: 'p' nulled but not freed upon failure  [memleakOnRealloc]

This isn't going to actually solve any memory leaks, it's just good practice for critical error (e.g. no memory error) handling.

Sometimes you want to see "1.5 minutes" (or "1.5M") intead of "90 seconds"
(or "90S")

I noticed some left over questionable use of ulong here in byte_count related
functions, so fixed that.

This macro hasn't done anything meaningful since we stopped using really old
versions of Borland compilers (and std libraries) where strerror() returned a
string terminated with a line-feed (\n) character.

Not at all approaching functional yet.

Also add OS_CPU @-code, to report what CPU architecture the OS thinks it is.

Fix the OS_VER @-code truncation bug from previous commit.

Yeah, the Windows block doesn't some unsafe concatenations still.

We don't have a use case negative byte values in strings anywhere that I can think of.

Created wrapper for MSFT _ui64toa_() - unsigned 64-bit integer to ASCII string.

Add Petabyte (actually, Pebibyte) support to byte_count_to_str()  and byte_estimate_to_str().

Removed 'B' suffix from byte_estimate_to_str() output for values < 1024. Callers may display this returned string followed by "bytes" (or similar) and thus the 'B' suffix is redundant in that case. If the caller needs a 'B' suffix, they can detect no suffix and add it.

This is a millisecond timer that tries to avoid floating-point
operations.

Now it should keep working after 24 days.

Per https://man7.org/linux/man-pages/man3/pthread_yield.3.html
       This call is nonstandard, but present on several other systems.
       Use the standardized sched_yield(2) instead.

Should fix #299.

Its a terrible idea to modify the return value of strerror() on any platform. strerror() can (and often does) return immutable string constants - don't try to modify that string even if it does end in trailing white-space (find another solution to that). This change only affects non-*nix builds since we were already doing the right thing for *nix.

Define a thread-safe/re-entrant version of strerror(): safe_strerror() which is just thin wrapper around strerror_s on Windows and strerror_r elsewhere.

we're at it.

definition of ARCHITECTURE_DESC in builds targetting ARM processors.

builds.

nanosleep was interrupted by a signal and the SLEEP parameter is > 1.
Previously, any signal wuold abort SLEEP() prematurely. A SLEEP(1) (yield)
behaves the same as before since 1 is a special value where a short-sleep
duration is fine/expected on occasion.

Hopfully this doesn't break the Borland or Watcom (?!) builds.

the number of decimal places in the resulting string.

Created byte_estimate_to_str(), much like byte_count_to_str(), except the
byte count is rounded to the nearest tera/giga/mega/kilo-byte
(technically tebi/gibi/mebi/kibi-byte, but ugh, I'm old-sk00l!).

We need this for MSVC2013 and MSVC2017 builds.

then a clarifying or modifying word (year[s], month[s], day[s], etc.).
parse_byte_count() and parse_duration() now support a space between the
numeric value (decimal digits) and the modifier. Hopefully this doesn't break
anything, but it makes parse_duration() compatible with the strings generated
with duration_to_vstr(). We might want a verbose version of byte_count_to_str()
in the future, so I figured it made sense to go ahead and add the white-space
skipping/parsing to prase_byte_count() now.