> Albert D. Cahalan writes:
> >
> > Ulrich Drepper writes:
> > >Richard Gooch <Richard.Gooch@atnf.CSIRO.AU> writes:
> >
> > >> aio_init() doesn't appear to be part of the standard. "POSIX.4
> > >> Programming for the Real World" doesn't mention it.
> > >
> > > It comes from Irix. Since it's optional to call this
> > > function I've implemented it.
> >
> > This hypocrisy is just incredible. You will leave it out of the
> > header file, won't you? It is not Standard(TM) you know.
> > It's only implemented in one foreign OS (not two, like llseek)
> > and was not part of Linux libc 5 (like llseek was). Maybe Irix
> > is now considered a Standard(TM), unlike SunOS and AIX.
>
> You said it.
He did.
Somebody said this thread is boring, but indeed it is not.
Let me tell you that being Standard(TM) about asynchronous IO and IO
completion in UNIX/POSIX planet is being wrong, in not-only my opinion.
So, the only way to have chance to become right is probably to forget
Standard in this domain.
About the inflammatory above, the flamed guy said "it's optionnal to
call this function" and if he implemented it as "optionnal", there is
neither Standard nor portability problems with this implementation,
providing that he also correctly documented it as "not standard" and
"optionnal". BTW, having an "optionnal" function in a header file is
not a big pollution compared to contents of current UNIXes header files.
About llseek() and seek64() stuff, I will say that I donnot trust any
software that claims to handle file of more than 2 GB on 32 bit
architectures and O/Ses, because it complexifies things in a way that
makes bugs very probables and I donnot care it is standard ot not.
Regards,
Gerard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu