Re: PATCH: Merge objc-improvements-branch to mainline

[Marcel Weiher]

Ziemowit Laski <zlaski@apple.com> wrote:

[request for blocks]

> Still, if you're willing to implement this feature, and do so in a way 
> that
> does not break existing stuff, I'd personally have no objections to 
> putting
> this in.

On a somewhat related note:

I actually have/had a patch for "wide-returns" that allows Higher Order
Messaging to work better with Apple's PPC runtime architecture.

It allows the compiler to skip sign-extending the return value of 
methods with sub-word returns, which is useful if you've used
reflective capabilities to redirect a message the compiler thinks
is returning a BOOL to actually return an 'id'.

AFAICT, it never is a problem because gcc actually always returns
a valid full register even for sub-word returns.

Marcel