`timespec_get` and `timespec_getres` are both C23 functions. They're
similar to the POSIX functions `clock_gettime` and `clock_getres`, but
both sets of functions define extra, exclusive features. I didn't
implement the extra features yet, partially because Redox doesn't seem
to support them yet either.
Our strptime implementation uses subformatting to handle specifiers that
expand to other specifiers. For example, '%T' expands to "%H:%M%:%S".
Relibc calls strptime again to handle '%T' by expanding it.
However, our strptime zeroes out struct tm which clobbers old values. If
strptime is called with a format string like "%D%t%T", the set values
for '%D' are clobbered.
Zeroing out the struct is extra, unneeded work as well. The user may
want to preserve old values. We don't read from struct tm so we don't
have to worry about invalid values. Finally, `musl` and `glibc` don't
clear the values either so we can default to their behavior.
realpath: Fixing undefined behaviour in the second test. If the call fails the resolved_name argument cannot be used for error checking because its state is undefined by SUSv2.
pipe: Changing the order of close and write error handling code. Errors in close could overwrite errno after write errors, returning incorrect error messages.
gmtime: Removed duplicate checks
Other fixes for fseek, rename, mktime, putwchar
This will allow us to redefine the exit function.
For example:
```
#define exit(code) { \
fprintf(stderr, "%s:%d: exit(%s) in function ‘%s’\n",
__FILE__, __LINE__, #code, __func__); \
_exit(code); \
}
```
Apparently gmtime was already implemented when I made localtime, so we had two different things written from scratch. We decided in the relibc channel of the Redox OS Mattermost chat to use my code, as it is more extensively tested and perhaps is clearer in how it works.
Anhad Singh