This patch to the Go frontend corrects the handling of negative zero when using untyped constants. Go's untyped constants are ideal floating point numbers, and don't have negative zero or infinity or NaN. The Go frontend was incorrectly representing the negative of zero as a negative zero, since it just uses the MPFR library. This was detectable via the standard function math.Float64bits. This patch corrects the problem. I added a test case to the master Go testsuite, which will be copied into the gccgo testsuite in due course. Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu. Committed to mainline and 4.7 branch.
Ian
diff -r afcf18f60884 go/expressions.cc --- a/go/expressions.cc Mon Apr 23 09:59:25 2012 -0700 +++ b/go/expressions.cc Mon Apr 23 13:42:46 2012 -0700 @@ -13620,7 +13620,13 @@ this->clear(); this->classification_ = NC_FLOAT; this->type_ = type; - mpfr_init_set(this->u_.float_val, val, GMP_RNDN); + // Numeric constants do not have negative zero values, so remove + // them here. They also don't have infinity or NaN values, but we + // should never see them here. + if (mpfr_zero_p(val)) + mpfr_init_set_ui(this->u_.float_val, 0, GMP_RNDN); + else + mpfr_init_set(this->u_.float_val, val, GMP_RNDN); } // Set to a complex value.
