As someone that maintains code on multiple platforms, I would call this a bug,
a bad design, and a major source of portability headaches. When I need high
precision math, I want to get the same results on all platforms, not to mention
exchanging binary data. Add this to the Endian problem and you've got lots of
fun. Not.
It's time x86/AMD64 got a 128bit FP clean math library and ditch the 80bit
throwback from the 8 bit PC days.
jgg
This message posted from opensolaris.org
_______________________________________________
opensolaris-code mailing list
[email protected]
https://opensolaris.org:444/mailman/listinfo/opensolaris-code
