Incorrect numeric_limits::max

[libbooze]

I could be wrong, but it seems to be a bug.

When I do:

--- a/include/boost/decimal/decimal64.hpp
+++ b/include/boost/decimal/decimal64.hpp
...
-    static constexpr auto (max)        () -> boost::decimal::decimal64 { return {9'999'999'999'999'999, max_exponent}; }
+    static constexpr auto (max)        () -> boost::decimal::decimal64 { return {9'999'999'999'999'999, max_exponent-16}; }

I get correct value:
max: 9.99999999999999900000e+384

with original code I get infinity.

by correct I mean what https://www.open-std.org/JTC1/SC22/WG21/docs/papers/2009/n2849.pdf specifies:

Expansion: an rvalue of type decimal64 equal to the maximum finite number that can
be represented by an object of type decimal64; exactly equal to 9.999999999999999 x 10^384 (there are fifteen 9's after the decimal point)

This is just for decimal64, others have same issue.

I think adding test that uses "excessive" precision(e.g. std::print("max: {:.30e}\n", nl::max()); ) to print numeric limits might be useful since I literally copy/pasted the value from PDF(
9.999999999999999) into find of my output window and I saw that after fix above output matches(all digits after those nines were zero and exponent matched).

[mborland]

I believe you are right. Will fix shortly.