From: Dirk Hohndel <hohndel@infradead.org>
To: Anthony Towns <aj@erisian.com.au>
Subject: Re: [PATCH] Add simplistic reimplementation of strcasestr to compat 
	library
In-Reply-To: <h2r87b3a4191004122259ofc26c84dk7ebc53c250ffb263@mail.gmail.com>
References: <m3eiikc7rc.fsf@x200.gr8dns.org>
	<h2r87b3a4191004122259ofc26c84dk7ebc53c250ffb263@mail.gmail.com>
Date: Tue, 13 Apr 2010 05:42:39 -0700
Message-ID: <m38w8rcym8.fsf@x200.gr8dns.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Cc: notmuch@notmuchmail.org
Precedence: list

On Tue, 13 Apr 2010 15:59:24 +1000, Anthony Towns <aj@erisian.com.au> wrote:
> On Tue, Apr 13, 2010 at 14:10, Dirk Hohndel <hohndel@infradead.org> wrote:
> > +/* the semantic here actually puzzles me:
> > + =C2=A0 how can haystack be const char * - yet the return value is cha=
r *
> > + =C2=A0 after all, it points to a sub-string of haystack... */
>=20
> Dunno if this is a question from the original source, but the answer

No, that was me being puzzled :-)

> if anyone's interested is probably because C doesn't have templates --
> you'd ideally like to have it treated as:
>=20
>     char *strcasestr(char *haystack, const char *needle);
>=20
> for when you're doing a search and replace on the needle (say), and:
>=20
>     const char *strcasestr(const char *haystack, const char *needle);
>=20
> for when you're doing a search for the needle in something you can't
> modify. But C isn't clever enough to let you say that with just one
> function (and no fancy #defines), so you have to drop some of the
> typechecking with the (char*) cast on the return value if you want to
> handle both use cases, without the compiler complaining about
> const->non-const conversions in otherwise correct code in one case or
> the other.

That makes sense. Thanks

/D

--=20
Dirk Hohndel
Intel Open Source Technology Center