<sys/ioctl.h> on msdos

<sys/ioctl.h> on msdos

[Dan Nicolaescu]

Does msdos have <sys/ioctl.h> ?
In a few places 
#ifndef MSDOS
is used to avoid including it.
Other places use HAVE_SYS_IOCTL_H
process.c includes it unconditionally in the non-msdos part.

It would be nicer to include it unconditionally everywhere, or to use
the same conditional everywhere.

Re: <sys/ioctl.h> on msdos

[Eli Zaretskii]

> From: Dan Nicolaescu <dann@gnu.org>
> Date: Wed, 17 Nov 2010 23:42:13 -0500
> 
> Does msdos have <sys/ioctl.h> ?

It does, but this header defines DOS ioctl stuff, not the Posix ioctl
stuff.  There's no emulation of the Posix ioctl functionality in the
standard C library used to build the DOS port.

> In a few places 
> #ifndef MSDOS
> is used to avoid including it.

Yes, because doing so pollutes the namespace with gobs of symbols that
could get in the way.

> Other places use HAVE_SYS_IOCTL_H
> process.c includes it unconditionally in the non-msdos part.

I see these places where sys/ioctl.h is included in Emacs:

  - in process.c -- not relevant for MSDOS and included unconditionally
  - in keyboard.c -- conditioned by MSDOS
  - in sound.c -- conditioned by MSDOS
  - in xterm.c -- only relevant for DOS if someone revives the old DOS
    port of Xlib, which probably won't happen; conditioned by
    HAVE_SYS_IOCTL_H
  - in systty.h -- conditioned by HAVE_SYS_IOCTL_H

> It would be nicer to include it unconditionally everywhere, or to use
> the same conditional everywhere.

I could arrange for the MSDOS port to not define HAVE_SYS_IOCTL_H, and
then we could use that everywhere.

Re: <sys/ioctl.h> on msdos

[Eli Zaretskii]

> From: Eli Zaretskii <eliz@gnu.org>
> Date: Thu, 18 Nov 2010 05:26:26 -0500
> Cc: emacs-devel@gnu.org
> 
> > It would be nicer to include it unconditionally everywhere, or to use
> > the same conditional everywhere.
> 
> I could arrange for the MSDOS port to not define HAVE_SYS_IOCTL_H, and
> then we could use that everywhere.

Should I do this?

Re: <sys/ioctl.h> on msdos

[Dan Nicolaescu]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Eli Zaretskii <eliz@gnu.org>
>> Date: Thu, 18 Nov 2010 05:26:26 -0500
>> Cc: emacs-devel@gnu.org
>> 
>> > It would be nicer to include it unconditionally everywhere, or to use
>> > the same conditional everywhere.
>> 
>> I could arrange for the MSDOS port to not define HAVE_SYS_IOCTL_H, and
>> then we could use that everywhere.
>
> Should I do this?

Not sure.
Are you sure that including sys/ioctl.h unconditionally has some bad effects?

Is HAVE_SYS_IOCTL_H currently defined?

If yes, then it seems that the only extra places to include sys/ioctl.h would be
keyboard.c and sound.c.

Re: <sys/ioctl.h> on msdos

[Eli Zaretskii]

> Cc: emacs-devel@gnu.org
> From: Dan Nicolaescu <dann@gnu.org>
> Date: Sat, 20 Nov 2010 02:39:50 -0500
> 
> Are you sure that including sys/ioctl.h unconditionally has some bad effects?

No, I'm not.  It's just good engineering practice.

> Is HAVE_SYS_IOCTL_H currently defined?

Yes.  It's defined by one of the system headers included by config.h
(after config.in is edited by msdos/sed2v2.inp).

> If yes, then it seems that the only extra places to include sys/ioctl.h would be
> keyboard.c and sound.c.

sound.c doesn't matter, since MSDOS does not define HAVE_SOUND (so we
can remove that part altogether from sound.c).

Are you planning on removing HAVE_SYS_IOCTL_H altogether and not
testing for it in `configure'?  Because if HAVE_SYS_IOCTL_H is to
stay, there could be no harm in undefining HAVE_SYS_IOCTL_H on MSDOS:
it will be in on of the msdos/ Sed scripts, not visible in any of the
Emacs sources.  We will just replace a couple of "#ifndef MSDOS" with
"#ifdef HAVE_SYS_IOCTL_H".

If you do want to remove HAVE_SYS_IOCTL_H, then I guess it would be
okay to remove the MSDOS conditions from the places that include
sys/ioctl.h, and see if anything breaks.

Re: <sys/ioctl.h> on msdos

[Dan Nicolaescu]

Eli Zaretskii <eliz@gnu.org> writes:

>> Cc: emacs-devel@gnu.org
>> From: Dan Nicolaescu <dann@gnu.org>
>> Date: Sat, 20 Nov 2010 02:39:50 -0500
>> 
>> Are you sure that including sys/ioctl.h unconditionally has some bad effects?
>
> No, I'm not.  It's just good engineering practice.
>
>> Is HAVE_SYS_IOCTL_H currently defined?
>
> Yes.  It's defined by one of the system headers included by config.h
> (after config.in is edited by msdos/sed2v2.inp).
>
>> If yes, then it seems that the only extra places to include sys/ioctl.h would be
>> keyboard.c and sound.c.
>
> sound.c doesn't matter, since MSDOS does not define HAVE_SOUND (so we
> can remove that part altogether from sound.c).
>
> Are you planning on removing HAVE_SYS_IOCTL_H altogether and not
> testing for it in `configure'?  Because if HAVE_SYS_IOCTL_H is to

Yes, given that we are including sys/ioctl.h in unconditionally in a
few places, it makes no sense to have configure test if it exists.

> stay, there could be no harm in undefining HAVE_SYS_IOCTL_H on MSDOS:
> it will be in on of the msdos/ Sed scripts, not visible in any of the
> Emacs sources.  We will just replace a couple of "#ifndef MSDOS" with
> "#ifdef HAVE_SYS_IOCTL_H".

> If you do want to remove HAVE_SYS_IOCTL_H, then I guess it would be
> okay to remove the MSDOS conditions from the places that include
> sys/ioctl.h, and see if anything breaks.

I can't test that, so it would be great if you could.

Re: <sys/ioctl.h> on msdos

[Eli Zaretskii]

> Cc: emacs-devel@gnu.org
> From: Dan Nicolaescu <dann@gnu.org>
> Date: Sat, 20 Nov 2010 23:40:39 -0500
> 
> > Are you planning on removing HAVE_SYS_IOCTL_H altogether and not
> > testing for it in `configure'?  Because if HAVE_SYS_IOCTL_H is to
> 
> Yes, given that we are including sys/ioctl.h in unconditionally in a
> few places, it makes no sense to have configure test if it exists.

Then go ahead and remove HAVE_SYS_IOCTL_H and the conditions to
include sys/ioctl.h.

> > If you do want to remove HAVE_SYS_IOCTL_H, then I guess it would be
> > okay to remove the MSDOS conditions from the places that include
> > sys/ioctl.h, and see if anything breaks.
> 
> I can't test that, so it would be great if you could.

Sure, that's what I meant.

Re: <sys/ioctl.h> on msdos

[Eli Zaretskii]

> Cc: emacs-devel@gnu.org
> From: Dan Nicolaescu <dann@gnu.org>
> Date: Sat, 20 Nov 2010 23:40:39 -0500
> 
> > If you do want to remove HAVE_SYS_IOCTL_H, then I guess it would be
> > okay to remove the MSDOS conditions from the places that include
> > sys/ioctl.h, and see if anything breaks.
> 
> I can't test that, so it would be great if you could.

I ran a few sanity checks and didn't see anything that broke as result
of this change.