Joey K Tuttle wrote: > I am a bit surprised that results from C library routines would > lead to reduced accuracy in a newer release of J...
Could be a new compiler, or a new patch level on the platform hosting the compiler. Could be that the old behavior in some way violated the C standard, and that the new behavior is, thus, "more correct". For example, there are various bits which tune the behavior of floating point operations, and perhaps the default has been changed or perhaps the interpretation of those bits has changed. Or, perhaps the new mechanism used by C is faster or more portable or involves less register pressure or whatever than the old system. Or, quite likely, it's something I've not even suggested, above. -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
