[1.34.0] hash_float_test / gcc-cygwin-3.4.4
Looking at: http://tinyurl.com/yotd5w I see no signs of any misconfiguration -- seems like hash library problem on cygwin. Would anybody be able to look into this? Thanks, Volodya
On Mon, 19 Feb 2007 17:36:53 -0000, Vladimir Prus <ghost <at> cs.msu.su> wrote:
I see no signs of any misconfiguration -- seems like hash library problem on cygwin. Would anybody be able to look into this?
It looks like cygwin's ldexp and frexp don't support long doubles. Working around that would be too complicated at this stage. So I'll mark up the failure and look into dealing with it in 1.35. Daniel
Daniel James wrote:
On Mon, 19 Feb 2007 17:36:53 -0000, Vladimir Prus <ghost <at> cs.msu.su> wrote:
I see no signs of any misconfiguration -- seems like hash library problem on cygwin. Would anybody be able to look into this?
It looks like cygwin's ldexp and frexp don't support long doubles. Working around that would be too complicated at this stage. So I'll mark up the failure and look into dealing with it in 1.35.
Yes, this is fine. Thank you! - Volodya
Daniel James wrote:
On Mon, 19 Feb 2007 17:36:53 -0000, Vladimir Prus <ghost <at> cs.msu.su> wrote:
I see no signs of any misconfiguration -- seems like hash library problem on cygwin. Would anybody be able to look into this?
It looks like cygwin's ldexp and frexp don't support long doubles. Working around that would be too complicated at this stage. So I'll mark up the failure and look into dealing with it in 1.35.
Right: there is no long double std lib support - or rather there is, but the arguments get truncated to double internally :-( Causes all kinds of traps for the unwary: the one I "loved" was sqrt returning +INF when passed a finite value! FYI BSD has the same issues. HTH, John.
participants (3)
-
Daniel James -
John Maddock -
Vladimir Prus