Re: inlining + optimization = nuisance bugs

> > [I've never understood why people are so eager to have fast wrong
> > answers. -John]


What are some good books on numerical methods for compiler writers?
Incidentally, I'm with John on this one; I do most of my serious
numerical work in languages that support extended-precision numerics,
just because I don't want 'em to crash when some number geos outside
what its "reasonable" range seems to be today.


It seems to me that implementing a language without taking advantage
of numeric optimizations and inlining is, well, silly. At the same
time, having them, and having them behave correctly, is a tricky
thing. Surely some references are available.


I'm also interested in the "utility" functions that compilers (or at
least math libraries) are expected to provide, such as the
trigonometric functions; how does one compute them to some arbitrarily
chosen precision?


Ray Dillinger
[I can tell you what the standard books were fifteen years ago, but let's
see if anyone has something more recent than Cody and Waite. -John]






--

Return to the comp.compilers page.
Search the comp.compilers archives again.