Re: bug 10491 was Re: More astonishing progress in japi scores

[Andrew Haley]

Sven de Marothy writes:
 > On Thu, 2004-10-07 at 18:01, Per Bothner wrote:
 > > That other issue is that system strtod may not be accurate.
 > > Note that parseDouble is required to return the *closest*
 > > double.  This is a difficult requirement to meet, and cannot be done
 > > by simple (i.e. traditional) implementations.
 > 
 > Which is also a good argument against re-implementing it, unless we have
 > to. Apparently the strtod.c file in gcj/classpath is not actually from
 > fdlibm but rather from newlib. 

That's right, it is.  I forgot.

The newlib implementation might be easy to translate into Java, but it
has a rather awkward licence.

Andrew.