Put scope blocks on a diet

[Alexandre Oliva]

On Nov 29, 2007, "Richard Guenther" <richard.guenther@gmail.com> wrote:

> Heh - but I claim differences in UIDs shouldn't change generated code ;)

Ah, yes, indeed, that makes sense.  Maybe once we fix the dependencies
on DECL_UIDs, we should have an option to introduce random noise in
UID increments, to be used for testing that this desirable property is
maintained, even if we decide to keep DECL_UIDs in sync.


It might be useful to do the same to TYPE_UIDs (I know Java will need
work to ensure same code for these), and to have garbage collection at
random points too.  Bootstrap testing with these randomization options
enabled would help expose latent bugs.

-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
FSF Latin America Board Member         http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}