Test cosh, sinh in non-default rounding modes (bug 3976)

[Andreas Jaeger]

On Friday, March 02, 2012 22:45:36 Joseph S. Myers wrote:
> The submitter of bug 3976 reports wild results from cosh and sinh in
> non-default rounding modes.  I can reproduce this with older glibc,
> but not after my exp patch (the problems presumably arose from exp as
> called by those functions).  This patch adds tests for cosh and sinh
> in each rounding mode (with arguments that showed problems with older
> glibc) to protect against this problem appearing again.
> 
> The error of 27ulp for long double sinh in round-upwards mode is
> larger than ideal (it becomes the largest checked-in ulps value; the
> largest was 24 after my recent change to reduce some values), but I
> don't think it's enough to merit a separate bug (or making sinh save
> and restore the rounding mode itself).

Thanks, this is fine.

Andreas
-- 
 Andreas Jaeger aj@{suse.com,opensuse.org} Twitter/Identica: jaegerandi
  SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
   GF: Jeff Hawn,Jennifer Guild,Felix Imendörffer,HRB16746 (AG Nürnberg)
    GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126