We can't really use things in std, since they're not available on all platforms. We had a bumch of #ifdef code for platforms which do support std::numeric_limits, but it was too difficult to maintain. At some point, when we phase out support for older compilers, we'll probably upgrade the code to use std::numeric_limits.
Dave
Morten Bach
M�ller To: <[EMAIL PROTECTED]>
<Morten@thydatac cc: (bcc: David N
Bertoni/Cambridge/IBM)
enter.dk> Subject: sqrt(-2.01)
04/08/2002 03:52
AM
Please respond
to xalan-dev
Hi,
In the DoubleSupport.cpp file there is a constant
const double DoubleSupport::s_NaN = sqrt(-2.01);
This assignment prevents me from using the PlatformSupport.DLL in an
executable since it throws a Float support exception - probably due some
strange way the application hosting the DLL interacts with the float
library.
It would be nice if the limits package could be used instead:
const double DoubleSupport::s_NaN =
std::numeric_limits<double>::infinity();
kind regards
Morten
