Re: [PATCH]: Promote float args on i386 (attn Hurd, Cygwin, DJGPP...)

[Michael Snyder <msnyder at cygnus dot com>]

Mark Kettenis wrote:
> 
> The differences were causing the problems you were seeing.  Looks like
> nobody noticed any problems since the compilers that we were using
> produced the correct debugging info and therefore GDB is never forced
> to do the default promotion.
> 
> Looks like on more reason to get rid of K&R C to me :-).

Can't do that, of course.  ;-(

But what would you think of defining COERCE_FLOAT_TO_DOUBLE?
If I understand correctly, it only comes into play in the
case where the compiler does not produce the correct debug
information.  Since apparently some do and some don't, 
that would be an improvement, right?

BTW it worked for Solaris-x86...