Date: Tue, 3 Jan 2012 19:20:20 +0100 From: Ed Schouten <ed@80386.nl> To: Bruce Evans <brde@optusnet.com.au> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r229368 - in head: lib/libc lib/libc/arm/string lib/libc/i386/string lib/libc/mips/string lib/libc/string lib/libstand sys/boot/userboot/libstand Message-ID: <20120103182020.GH1895@hoeg.nl> In-Reply-To: <20120104013401.S6960@besplex.bde.org> References: <201201030714.q037E2qq010125@svn.freebsd.org> <20120104013401.S6960@besplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Hi Bruce,
* Bruce Evans <brde@optusnet.com.au>, 20120103 16:49:
> This breaks the Standard C namespace. When they are in the same object
> file, there is no way to get the standard name without getting the
> nonstandard name. So the following C-standard-conforming C program
> now gets a linkage error (multiple definition of `index'), at least with
> static linkage:
>
> #include <string.h>
> int index;
> void foo(const char *p) { return strchr(p, '1'); }
Though I sympathize, this problem is not just limited to strchr(). There
are other portions of the C library that use index() as well. For
example, if you use <ttyent.h>, ttyslot(), one of the exec*() functions
or the NIS functions, you already get index() linked into your binary.
> [...] and the C
> standard might require memcpy and memmove to have different addresses.
I just skimmed through the standard, and if I haven't overlooked
anything, no such requirement is made. Also, I can imagine a compiler
with good support for link-time optimisation can already merge equal
pieces of code together, making it even harder to reason about
inequality of function addresses.
Still, I am willing to address the issues you raised. index() and
rindex() aren't that important nowadays and I have a patchset ready in
my home directory that converts almost all apps in the base system to
use strchr() anyway.
As I don't feel like polluting the MI strchr() implementations with
index()/rindex() support, would it be okay if I implement index() and
rindex() as simple C functions that call into strchr() and strrchr()?
--
Ed Schouten <ed@80386.nl>
WWW: http://80386.nl/
[-- Attachment #2 --]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (FreeBSD)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=qvDX
-----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120103182020.GH1895>