Re: [PATCH] ioctl.2: note "int request" form, HISTORYise a bit

[Alejandro Colomar <colomar.6.4.3@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 3300 bytes --]

Hi наб,

On 2023-09-04 16:29, наб wrote:
> Bit me in https://github.com/thecoshman/http/issues/155:
> musl (and, thus, bionic) takes an int!
> 
> Kill the "POSIX says 'request' is int" comment,
> POSIX says nothing about ioctl:

Please reword that; while it will say nothing soon, the current
standard is Issue 7, which still says.  Something like your reply
to Jakub should be enough.

> it invented tc[gs]etattr() and tc[gs]etwinsize() to avoid having
> ioctl() at all, and STREAMS is long dead from POSIX.
> 
> Note how the interface evolved in HISTORY. The NetBSD CVS has
>   revision 1.1
>   date: 1993-03-21 10:45:37 +0100;  author: cgd;  state: Exp;
>   branches:  1.1.1;
>   Initial revision
>   ----------------------------
>   revision 1.1.1.2
>   date: 1995-02-27 11:49:43 +0100;  author: cgd;  state: Exp;  lines: +6
>   -6;
>   from Lite
>   ----------------------------
>   revision 1.1.1.1
>   date: 1993-03-21 10:45:37 +0100;  author: cgd;  state: Exp;  lines: +0
>   -0;
>   initial import of 386bsd-0.1 sources
>   ----------------------------
> and it's char * in 1.1.1.1 and 1.1.1.2 but ... in 1.1, so hell knows.
> 
> Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
> ---
>  man2/ioctl.2 | 40 ++++++++++++++++++++++++++++++++++++----
>  1 file changed, 36 insertions(+), 4 deletions(-)
> 
> diff --git a/man2/ioctl.2 b/man2/ioctl.2
> index 6b55d47c9..a658da354 100644
> --- a/man2/ioctl.2
> +++ b/man2/ioctl.2
> @@ -20,9 +20,8 @@ .SH SYNOPSIS
>  .nf
>  .B #include <sys/ioctl.h>
>  .PP
> -.BI "int ioctl(int " fd ", unsigned long " request ", ...);"
> -.\" POSIX says 'request' is int, but glibc has the above
> -.\" See https://bugzilla.kernel.org/show_bug.cgi?id=42705
> +.BI "int ioctl(int " fd ", unsigned long " request ", ...);" "\fR  /* glibc, BSD */\fP"
> +.BI "int ioctl(int " fd ", int " request ", ...);" "\fR            /* musl, other UNIX */\fP"

LGTM.

>  .fi
>  .SH DESCRIPTION
>  The
> @@ -103,7 +102,40 @@ .SH VERSIONS
>  .SH STANDARDS
>  None.
>  .SH HISTORY
> -Version\~7 AT&T UNIX.
> +Version\~7 AT&T UNIX has
> +.nf
> +.ti +.5i

I prefer using man(7) macros.  Please use

.PP
.in +4n
.nf
stuff here
.fi
.in
.PP

If you want to prevent the blanks that this adds, you can
enclose that in .PD 0 / .PD

(.in is not a man(7) macro, but it is already widespread.)

Cheers,
Alex

> +.BI "ioctl(int " fildes ", int " request ", struct sgttyb *" argp );
> +.fi
> +(where
> +.B struct sgttyb
> +has historically been used by
> +.BR stty (2)
> +and
> +.BR gtty(2),
> +and is polymorphic by request type (like a
> +.B void *
> +would be, if it had been available)).
> +.PP
> +SysIII documents
> +.I arg
> +without a type at all.
> +.PP
> +4.3BSD has
> +.nf
> +.ti +.5i
> +.BI "ioctl(int " d ", unsigned long " request ", char *" argp );
> +.fi
> +(with
> +.B char *
> +similarly in for
> +.BR "void *" ).
> +.PP
> +SysVr4 has
> +.nf
> +.ti +.5i
> +.BI "int ioctl(int " fildes ", int " request ", ... /* " arg " */);"
> +.fi
>  .SH NOTES
>  In order to use this call, one needs an open file descriptor.
>  Often the

-- 
<http://www.alejandro-colomar.es/>
GPG key fingerprint: A9348594CE31283A826FBDD8D57633D441E25BB5

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]