K 10
svn:author
V 5
kargl
K 8
svn:date
V 27
2015-03-10T17:10:54.781845Z
K 7
svn:log
V 350
According to POSIX.1-2008, the Bessel functions of  second kind
should raise a divide-by-zero floating point exception for x = +-0
and an invalid floating point exception for x < 0 including x = -Inf.
Update the code to raise the exception and update the documentation
with hopefully better description of the behavior.

Reviewed by:	bde (code only)

END