Re: Numeric types, overhead and precision loss

[David Kastrup <dak@...>]

"Grellier, Thierry" <t-grellier@ti.com> writes:

> if you need 64 bits int (well > 53 bits).

Then an integer expanding into float will not help.  You'll need an
arbitrarily sized integer instead.  Which would make more sense as a
separate data type since it would not at some point of time start
bleeding precision.

Still, it would appear to me that the main motivation to have integers
as a separate data type would be to be able to pass arrays of integers
(or similar) without reinterpretation between external code and Lua.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum