Eclipse vs gij/libgcj, unit-at-a-time, -O2 and -O

[Andrew Haley]

Jeff Sturm writes:
 > On Mon, 3 Nov 2003, Andrew Haley wrote:
 > > Mark Wielaard writes:
 > >  >
 > >  > Jeff said that there were already bug reports for inlining issues
 > >
 > > Which ones?  IMO these should be critical.
 > >
 > > I only see
 > >
 > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11533
 > 
 > You found the one I was thinking of.  I thought "critical" was meant
 > for regressions... but I guess this is a regression if it is now
 > happening at -O2 (the PR mentioned -O3).

Well, I'm not sure how an inlining problem can happen at -O2.

However, a failure in the tree inliner that worked with the RTL
inliner is IMO a regression.

Andrew.