This is the mail archive of the
mailing list for the GCC project.
Re: Patch to gfortran/13742
I think this patch is responsible for PR 15641. The testcase from the PR
works with a gfortran compiled before your change (May 19th), and the
only ChangeLog entry which seems to affect this is this one.
Paul Brook wrote:
Your patch assumed that the items in a segment were sorted. I've modified you
patch so that condition is maintained while building the segment.
I also tidied up the way we built the equivalence chains. The algorithm is
still O(something nasty), but hopefully a much smaller O ;)
Tested on i686-linux. Applied to mainline.