Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 7 Apr 2007 21:06:50 +0400
From:      Andrey Chernov <ache@FreeBSD.org>
To:        Pawel Jakub Dawidek <pjd@FreeBSD.org>
Cc:        cvs-src@FreeBSD.org, src-committers@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/include stdio.h src/sys/sys unistd.h
Message-ID:  <20070407170650.GA10631@nagual.pp.ru>
In-Reply-To: <20070407165435.GK63916@garage.freebsd.pl>
References:  <200704071602.l37G2V1c066806@repoman.freebsd.org> <20070407164022.GA10309@nagual.pp.ru> <20070407165435.GK63916@garage.freebsd.pl>

next in thread | previous in thread | raw e-mail | index | archive | help

--+QahgC5+KEYLbs62
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sat, Apr 07, 2007 at 06:54:35PM +0200, Pawel Jakub Dawidek wrote:
> On Sat, Apr 07, 2007 at 08:40:22PM +0400, Andrey Chernov wrote:
> > On Sat, Apr 07, 2007 at 04:02:31PM +0000, Pawel Jakub Dawidek wrote:
> > >   - Only define SEEK_DATA and SEEK_HOLE in sys/unistd.h when neither
> > >     _POSIX_SOURCE nor _XOPEN_SOURCE is defined.
> >=20
> > 1) This new addition should be exluded for !define(_C99_SOURCE)
> > !define(_ANSI_SOURCE) !define(_POSIX_C_SOURCE) too.
> >=20
> > 2) We design our *_VISIBLE framework right for the reason to not mentio=
n=20
> > all possible *_SOURCE each time like this, making includes unnecessary =
big=20
> > and hard to adapt to the future *_SOURCE tags, but mention one *_VISIBL=
E=20
> > tag instead, so please rewrite this thing using it.
>=20
> That's why I asked for help. _CDDL_VISIBLE is not good, because it is
> not related to CDDL license. _ZFS_VISIBLE is not good, because it is not
> ZFS-specific. _{SUN,SOLARIS,OPENSOLARIS}_VISIBLE is also not good,
> because it is already in Linux too. Solaris simply defines
> __EXTENSIONS__. Maybe we need something like this? I don't think we need
> separate _*_VISIBLE defines for every new #define in unistd.h and other
> headers, so something similar to __EXTENSIONS__ makes sense to me.

Ever for simple few defines *_VISIBLE framework is better because it is=20
easily extensible to new possible *_SOURCE comes without needs to edit=20
_all_ includes where namespaces user after that (i.e. without adding new=20
!defined(_XXX_SOURCE) to all there).

__EXTENSIONS__ scope is too large and don't specific to be meaningful.
Why _{SUN,SOLARIS,OPENSOLARIS}_VISIBLE isn't good? If it is SOLARIS=20
permanent addition, it and possible other ones from there perfectly fits=20
IMHO.

The patch below using __ZFS_VISIBLE as example, feel free to change to=20
better name.

--- cdefs.h.bak	Sat Sep 23 18:59:06 2006
+++ cdefs.h	Sat Apr  7 20:51:34 2007
@@ -492,6 +492,7 @@
 #undef _POSIX_C_SOURCE
 #define	_POSIX_C_SOURCE		199506
 #endif
+#define	__ZFS_VISIBLE		0
 #endif
=20
 /*
@@ -521,6 +522,7 @@
 #define	__POSIX_VISIBLE		198808
 #define	__ISO_C_VISIBLE		0
 #endif /* _POSIX_C_SOURCE */
+#define	__ZFS_VISIBLE		0
 #else
 /*-
  * Deal with _ANSI_SOURCE:
@@ -538,16 +540,19 @@
 #define	__POSIX_VISIBLE		0
 #define	__XSI_VISIBLE		0
 #define	__BSD_VISIBLE		0
+#define	__ZFS_VISIBLE		0
 #define	__ISO_C_VISIBLE		1990
 #elif defined(_C99_SOURCE)	/* Localism to specify strict C99 env. */
 #define	__POSIX_VISIBLE		0
 #define	__XSI_VISIBLE		0
 #define	__BSD_VISIBLE		0
+#define	__ZFS_VISIBLE		0
 #define	__ISO_C_VISIBLE		1999
 #else				/* Default environment: show everything. */
 #define	__POSIX_VISIBLE		200112
 #define	__XSI_VISIBLE		600
 #define	__BSD_VISIBLE		1
+#define	__ZFS_VISIBLE		1
 #define	__ISO_C_VISIBLE		1999
 #endif
 #endif
--- unistd.h.bak	Sat Apr  7 20:02:30 2007
+++ unistd.h	Sat Apr  7 20:56:06 2007
@@ -108,7 +108,7 @@
 #define	SEEK_CUR	1	/* set file offset to current plus offset */
 #define	SEEK_END	2	/* set file offset to EOF plus offset */
 #endif
-#if !defined(_POSIX_SOURCE) && !defined(_XOPEN_SOURCE)
+#if __ZFS_VISIBLE
 #define	SEEK_DATA	3	/* set file offset to next data past offset */
 #define	SEEK_HOLE	4	/* set file offset to next hole past offset */
 #endif



--=20
http://ache.pp.ru/

--+QahgC5+KEYLbs62
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.3 (FreeBSD)

iD8DBQFGF8+qVg5YK5ZEdN0RAqz2AJ9Oz/K6xrQccPlb/P5IfsHdeiM8iwCdFYuZ
XECZOH4vV3afD2Rpu/Trih0=
=KhIN
-----END PGP SIGNATURE-----

--+QahgC5+KEYLbs62--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070407170650.GA10631>