Farid Zaripov wrote:
I see that the in ldbl_denorm_min() used DBL_MIN constant.
Is this by design, or a bug?
I suspect it's a benign bug (copy and paste typo).
Below is a patch to make ldbl_denorm_min() consistent with
dbl_denorm_min() and flt_denorm_min().
Shouldn't the value be set to LDBL_MIN instead of 0? And has_denorm
to the result of denorm_min() < min() instead of denorm_min() != 0?
Martin
Index: etc/config/src/INFINITY.cpp
===================================================================
--- etc/config/src/INFINITY.cpp (revision 629828)
+++ etc/config/src/INFINITY.cpp (working copy)
@@ -720,8 +720,8 @@
for (volatile long double tmp = 1.0L; tmp; tmp /= 2.0L)
val.val = tmp;
- if (DBL_MIN < val.val)
- val.val = DBL_MIN;
+ if (LDBL_MIN <= val.val)
+ val.val = 0.0L;
#else
Farid.
