Re: Tracking performance regressions in 1.1.7-pre

On Saturday 16 October 2004 08:18, Grzegorz B. Prokopski wrote:
> On Fri, 2004-10-15 at 13:11, Michael Koch wrote:
> > On Friday 15 October 2004 18:13, Laurent Martelli wrote:
> > > >>>>> "Michael" == Michael Koch <[email protected]> writes:
> > >
> > > [...]
> > >
> > >   Michael> GCJ 3.3.x and up work nicely.
> > >
> > > I had gcj 3.3.4.
> >
> > Same here.  In pure GNU classpath upstream this works. I build it
> > on a daily basis here with GCJ 3.3 and 4.0. Must be something
> > introduced by the changes from the SableVM developers.
>
> Possibly.  But, you know, if it works with "picky" java compiler
> like jikes...  I don't think jikes is broken.
>
> Anyway, we need something that is guaranteed to work.  Some people
> may have even older gcj versions installed on their machines, so
> we'd rather give clear signal that you *have to* use jikes to
> compile sablevm-classpath, than to let users suffer from strange
> errors like that one.

You are free to provide a patch ...

Its clearly documented that older GCJ don't work and I don't got a bug 
report yet that someone had problems with this.


Michael