Clarification on my comment w/1.0b3 on the netwinder

[Alexandre Oliva]

On Dec 10, 1998, Godmar Back <gback at cs.utah.edu> wrote:

>  I copied the dynamic array trick from the Sparc sysdepMethod.

Wow!  When I wrote it, I never thought it would be used in so many
ports... :-)

> I have never understood just why this works, and would not at all
> be surprised if it broke in egcs.

It didn't, at least not on SPARC.  I wouldn't expect it to break
gratuitously any other platform either.

> Maybe they changed the way they do dynamic arrays?

On SPARC, they worked exactly because of that: the dynamic array would
be stored just one word after the storage reserved for register
spilling in the stack frame for called methods, exactly where
additional arguments were supposed to be stored.  Maybe whoever
designed the dynamic array feature had something like what we do in
mind already...

> Kiyo reported at one point that it would work without optimization,
> but not when compiled with optimization.  I forgot what architecture
> that was for.

I think it was m68k

-- 
Alexandre Oliva  http://www.dcc.unicamp.br/~oliva  aoliva@{acm.org}
oliva@{dcc.unicamp.br,gnu.org,egcs.cygnus.com,samba.org}
Universidade Estadual de Campinas, SP, Brasil