Re: *ping* [PATCH] unit-at-a-time for java

[Andrew Haley <aph at redhat dot com>]

Jeff Sturm writes:
 > On Wed, 3 Sep 2003, Andrew Haley wrote:
 > >  > The following patch (together with
 > >  > http://gcc.gnu.org/ml/gcc-patches/2003-09/msg00082.html) corrects both and
 > >  > enables unit-at-a-time for gcj.  It avoids the problem I reported earlier
 > >  > (ICE in dwarf2out.c) by defining a start_inlining hook that cancels
 > >  > inlining if the function is not yet expanded.
 > >
 > > Eeeeh.  So we only inline in order?  Nasty.  Still, I guess it's the
 > > best we can do at the present time.
 > 
 > It's not as terrible as it sounds.  cgraph_postorder generally ensures
 > called functions are expanded first, so inlining now works regardless of
 > declaration order.

Oh, right, that's great.

 > Also, gcj can now inline constructors.  For whatever reason these didn't
 > have DECL_INLINE set before.


I guess soon we'll want to find a way to control inlining on a
method-by-method basis.

Andrew.