This is the mail archive of the
mailing list for the GCC project.
Re: fix middle-end/19515
On Jan 29, 2005, at 6:18 AM, Ranjit Mathew wrote:
Richard Henderson wrote:
The solution is about what I mentioned in the PR. At each step, see
if we've got an incompletely initialized union, and if so, force the
entire object to be zeroed. Not spectacularly efficient, but
to avoid the quadratic match-up-the-initializer-with-the-structure
problem, which we should still solve at some point. But not now.
I'm seeing a bootstrap failure (i686-pc-linux-gnu). Not
completely sure that it's caused by your patch per se, but
the faulting code is what is touched by your patch:
This was filed under PR 19687 by the way with a reduced source and