Re: PATCH: minor C++ front end compile speed improvement

[Fariborz Jahanian <fjahanian at apple dot com>]

On Dec 1, 2004, at 3:28 PM, Matt Austern wrote:

> On Dec 1, 2004, at 2:44 PM, Fariborz Jahanian wrote:
>
> > On Dec 1, 2004, at 2:38 PM, Matt Austern wrote:
> > > No!  It reduces one branch with no branch.  We're already checking 
> > > to see whether this variable is NULL.  We're not adding any new 
> > > tests.
> >
> > I do not have the patch handy, But macro has a branch in it.
> >
> > #define ORIGINAL_NAMESPACE(NODE)  \
> >   (DECL_NAMESPACE_ALIAS (NODE) ? DECL_NAMESPACE_ALIAS (NODE) : (NODE))
> >
> > And you are putting an if-then-else around it.
>
> No.  The "if" was already there.  All I'm doing is adding an else 
> clause, so that we don't use this macro in cases where we know for a 
> fact it will do nothing.

The else clause causes a branch around the macro for cases where we 
know that macro does noting.

- fariborz

> 			--Matt