Notes from the PSF review are here: DM/SAT/DR-20100430-PSF.

Other notes:

Probably need to have a TCT discussion of the use of compiler-dependent !#ifdefs.

Gregory suggests policy that user code not be permitted to use compiler #ifdefs to change what code actually is compiled - instead, compiler dependencies would be hidden by wrapper functions or, in extremis, macros centrally defined.

Robert and Mike point out that !#pragma (warning ...) types of code may be unavoidable. This is generally OK with Gregory. Would still prefer to limit their use and prefer centralized disabling of widely annoying warnings. Some examples given of edge cases where that doesn't work.

Need future TCT and maybe SAT discussion of this. Need to determine who would formulate a proposal to the TCT. Probably Gregory.

Mike will contribute a platform-independent wrapper for isnan().