RE: [PATCH][libcpp] New deferred output pragma

["David Stubbs" <stubbs at IceraSemi dot com>]

> But I also would like more detailed discussion of why this patch
> &
> deferred_output pragma is useful. Could the original poster
> (Devid
> Stubbs?) give some realistic uses cases? Why does he want the
> patch?

We use this patch in combination with the patch posted by my colleague
Rahul (http://gcc.gnu.org/ml/gcc-patches/2010-04/msg00581.html). 

We have some assert macros that previously expanded to a lot of inline
code. We now generate a thunk for these macros to reduce the amount of
inline code, and use section attributes on the generated function to
place it in a less expensive memory segment. 

I can provide a fuller example if necessary, but it's quite a unique use
case. 

Are there any other problems with the patch, aside from the ones Tom
pointed out? I'll happily re-work the patch if deemed useful.

Thanks!
Dave.