Re: Phix 0.8.3 uploaded
- Posted by petelomax Feb 02, 2021
- 2092 views
irv said...
the result of 7 * 8.50 is: 3.84062e+248./p -test hangs here.
Don't know much about math, but that seems to be off by a few percent
Oh yes, I forgot about that - it's not a "real" bug.
Remember that C_DOUBLE stuff? Well, t42cback lines 387-437 (which I sidestepped at the time) are now effectively a test it is still working wrong.
In other words it is a hand-crafted inline assembly test that is based on my faulty understanding of lnx 64-bit calling conventions.
Option a: completely rewrite it and perhaps/probably still be wrong, or b: bin it. The latter is going to win.