What to do about widely used ioctl.scala?

[LeeTibbert]

To ensure that everything implemented in posixlib was documented in ReadTheDocs, I did a sweep through
comparing the documentation with the contents of posixlib.

I discovered a blivet when it came to ioctl.scala. It needs a new home
and to be removed from the posixlib documentation.

It is not POSIX. Posix had an ioctl but it was in strops.h
and, to boot, is now marked obsolescent ( their word, English
I would use, my grammar teacher would blanch)

It is not part of the ISO C standard library, under any name I could find.

Now the delicate part is that posix.sys.ioctl (sic) is used in a number
of places in Scala Native. I suspect that it is also used in the wider
SN world. If it is going to change location, and I believe it should,
the breaking 0.5.0 is the time to do it.

I have talked often about the need in SN for a "WYSIWYG" (what you
see is what you get) or OS specific (gnu, Linux, MacOS, (Windows?))
library. This is another case, although it would probably go in a "shared"
sub-directory. There are several other candidates for such an "Wild West"
non-POSIX, non-ISO_C library. It is kinda like the case of 'irregular verbs'.

I invite discussion. Let's see if we can get some consensus.

There is a hack solution, but I do not want to document that in public.
We have enough of these "special" cases to drive an architectural (project)
solution.

[ekrich]

My guess is that it is located in POSIX because of this - https://man7.org/linux/man-pages/man2/ioctl_tty.2.html

You can also find it as a BSD API which makes since because it would have to compile on macOS too.
https://www.freebsd.org/cgi/man.cgi?sektion=2&query=ioctl

I would say "obsolescent" or not that strops.h would be ok if FIONREAD was defined in POSIX but it isn't so that isn't a good solution.

I found this so it even makes it more unclear where ioctl should land or if there is a POSIX replacement.
https://stackoverflow.com/questions/7294732/replacement-for-the-ioctl-function