I was rather glad when I first encountered DEC64 because I had previously had the displeasure of having to think deep thoughts about IEEE 754. DEC64 seemed simple enough that I could get the gist on a note card, and generally I gravitate towards the simplest possible options whenever possible.
However, reading some of the comments on previous HN discussion of DEC64 are kind of depressing. It sounds kind of like either you get a floating point number representation that's comprehensible or you get one that's fast (or you get neither ... which sounds like it might be the case for DEC64).
Are my options really some sort of symbolic representation OR IEEE 754? Bleh that seems rather unfortunate.
However, reading some of the comments on previous HN discussion of DEC64 are kind of depressing. It sounds kind of like either you get a floating point number representation that's comprehensible or you get one that's fast (or you get neither ... which sounds like it might be the case for DEC64).
Are my options really some sort of symbolic representation OR IEEE 754? Bleh that seems rather unfortunate.
(Side note: someone linked this on the previous discussion ... I was surprised http://en.wikipedia.org/wiki/Decimal64_floating-point_format)