hexadecimal/binary floating point literals

[gavinking]

I wonder if we should after all support floating point literals in bases other than 10. I guess this can be useful for some very technical floating-point stuff, like, for example:

Float minInteger = -$1.0p63;
Float maxInteger = $1.0p63 - 1;
Float minFloat = $1.0p-1022;
Float maxFloat = ($2.0 - $1.0p-52) * $1.0p+21023;

I note, however, that I have never once, in my whole career, used such a thing nor seen it used. Perhaps if we ever decide to implement ceylon.math in Ceylon, we might find one or two places to use it?

Anyway, if anyone knows of a use for these, please let me know.

[FroMage]

I don't think that's useful at all. If anything, I suppose that people who want floating point literals in binary are going to want to type them in binary itself (that is, in the binary encoding for floating point, not in the binary base). I've never heard of floating points in hexa.

[gavinking]

Java supports this. That's the only reason I raised it.

[FroMage]

That's interesting. We should unearth the discussion that led to them supporting it, if it exists.