Re: RFA: strtod patch

[Per Bothner <per at bothner dot com>]

Mark Mitchell wrote:

> Fine; I wasn't sure where the Java-specific checking was supposed to go.

At this point there is no attempt to have a generic strtod, so we might 
as well
do syntax checking in a Java-specific way.

However, if for example glibc had a guaranteed-accurate implementation of
strtod, it might be different.  I don't know if glibc does.  (I believe 
doing so
correctly requires using bignums, at least in some cases, so it is 
non-trivial.)
In that case, it might be nice to try to autoconf test for witha good 
strtod version
and use it if available, perhaps with a backup version in xibiberty. 
 (libstdc++
could use the same version.)  And in that case we need to do Java syntax 
checking
at a separate layer.