This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: kernel-2.2.1-undefined references.
- To: egcs at egcs dot cygnus dot com
- Subject: Re: kernel-2.2.1-undefined references.
- From: Bob McWhirter <bob at werken dot com>
- Date: Mon, 8 Feb 1999 19:59:10 -0500 (EST)
On Mon, 8 Feb 1999, Linus Torvalds wrote:
> On Mon, 8 Feb 1999, Lee Iverson wrote:
> >
> > It occurs to me that we do, in fact have an acceptable method of
> > *guaranteeing* inlining of a *function* in gcc. That is a combination
> > of a macro and the "Statement Expression" extension.
>
> Sure.
>
> Why don't you just remove "inline" completely, then?
I'm probably coming into this waaaay too late, but I thought the
standards (C++ and C9X?) had definite things to say about what
'inline' should mean; that is, that it was a hint or recommendation
to the compiler, but by no means a direct imperative.
It seems that a guaranteed forced 'inline' would be non-conforming.
In that case, could there not be a magical Linus-Special Inline
such as __inline or otherwise, that'd be Just Like inline, but
Better(tm)?
As far as egcs breaking things, I thought we pretty much expected
that (it's half the fun), and rely on gcc2 if we needed actual
determinism. ;)
Feel free to ignore.
-Bob
--
Bob McWhirter Shop Foreman, Decision Scientist, Cat Juggler
Werken Digital A division of the Benjamin J. Werken & Sons Company
bob@werken.com http://www.werken.com./