I've come up with my craziest idea yet, which actually turned
out to have a fairly straightforward implementation (so far).
It packs integers and floating-pointer numbers semi-indirectly
into an integer "token", and can be extended to support
additional types such as rationals and extended-precision
integers.
The idea is described in more detail in this comp.lang.apl post
Subject: Representation of numbers 4/5/2015
https://groups.google.com/d/topic/comp.lang.apl/KSwo40-eLKI/discussion
I'd appreciate "conceptual" comments to follow-up in that
thread.
But I'd love to hear any thoughts on the implementation,
which is in C, and the logical place is here where I'm telling
you about it.
The implementation so far is represented by the latest commit.
It's not fully tested, but compiles (gcc no args)
and reads ok (to me).
https://github.com/luser-dr00g/inca/commit/16ba509a0b151c285321eab9c378967da4210286