Kill hack in java disabling unit-at-a-time

[Steven Bosscher]

On Monday 15 August 2005 12:50, Andrew Haley wrote:
>  > OK, for the very moment I will add the flag so we can at least make
>  > the testing easier.  For 4.2 however there are definite plans for
>  > dropping non-unit-at-a-time (there are still problem with few
>  > specific sources in glibc and runtime but at very minimum I plan to
>  > drop non-unit-at-a-time inliner) so we need to have this resolved
>  > by then.
>
> Well, that's assuming it's possible to solve the explosive memory
> growth problem.

Wasn't that for the most part also a front-end problem?
IIRC we had a discussion about that not too long ago on
IRC, but I don't recall the details -- but wasn't there
something about gcj that makes it work in whole-program
mode??

Oh, and it'd help if the java front end would stop using
TREE_COMPLEXITY and TREE_CHAIN on expression nodes ;-)

Gr.
Steven