occur (for example, -march=core2 used to break build). Not a FreeBSD
problem even -- Linux crowd was bitten by this too (and misdiagnosed
it). Our package-building does not include the flag, which is why
we haven't seen this before.
Attempt to help configure better detect the architecture and
CPU-capabilities -- and use assembler-implementations of various
routines, where possible. ("make check" is almost 30% faster now)
Ensure, SSE2-specific code builds properly with clang as well as gcc.
Ensure, the author's "make check" runs as regression-test.
Add another patch found on SourceForge.
(Attempt to) unbreak on PowerPC -- untested.
Bump PORTREVISION. Take maintainership for the time being, to deal
with any fallout.