🚀 go-pugleaf

> While there may be other problems, I find 2 errors in <float.h> supplied
> with my copy of gcc.  FLT_DIG is reported as 6 where 7 is correct. 

You don't give what CPU type your machine is, so I'll assume it uses
IEEE754 single precision for floats.  If this is true, 6 is the
correct value.

The number of decimal digits is computed as follows:

	floor( (p-1) x log10 b) + (b == power of 10)

	= floor((24-1) x log10 2) + 0 

	= floor(23 x 0.301030...)

	= floor(6.923690...)

	= 6

IEEE singles have almost but not quite 7 decimal digits of precision.

> This goes way back to the original proposal, so it may be hard to
> get it changed in the distribution.  DBL_MAX is given to only 16
> digits, where 17 are required to obtain the correct binary value.
> The correct value is 1.7976931348623157e308.  The value rounded of
> to 16 digits compiles to Infinity and that is not the fault of the
> compiler! 

Don't know about this one --- it's correct in my gcc installation.

	--jtc