Re: inline (Was: Bloat)

[Andrew Reilly <andrew-netbsd%areilly.bpc-users.org@localhost>]

On Sun, Feb 08, 2009 at 01:51:46PM +0000, David Holland wrote:
> On Sun, Feb 08, 2009 at 01:21:20PM +0000, David Brownlee wrote:
> >     Given you've been playing with it - do you know if there is
> >     any way around the issue Thor commented on - functions not
> >     being inlined when called from other compilation units?
> 
> In the absence of a global optimizer, there has to be an inline
> definition available in the translation unit in order for inlining to
> be done. So only inline functions defined as such in header files are
> reasonable candidates.

I've always thought that global optimization is the only
sensible way to let your abstraction levels integrate properly.
In particular, a global optimizer should automatically inline
any functions that are only called once irrespective of how
large they are.  Only trouble with the C linkage model is that
you can't know that unless you do global compilation, which is
(mostly) incompatible with things like dynamically loadable
modules.  It's a worry.

Anyway, the main point of this ramble is that I've noticed that
gcc on my Mac has an extra option (--combine, I think) that
gives some of the effect of global optimization.  Since it's in
gcc, maybe it's also available up-stream by now?

Cheers,

-- 
Andrew