[cxx-abi-dev] Request for clarification: floating point literal mangling

[Mark Mitchell]

> I don't know what the existing implementations have done, but I suspect
> that this is a misstatement and that we meant without trailing zeroes,
> which makes much more sense for FP.  But at this point, the confusion
> should be resolved based on what the implementations do.

The current GCC implementation is just plain wrong.

The current EDG implementation is close -- but doesn't try to skip any
leading zeros.  It does high-order bytes first, using the target
representation.

HP?  IBM?

I suspect that the EDG implementation is what we want.

-- 
Mark Mitchell
CodeSourcery, LLC
mark at codesourcery.com