Re: Hexadecimal floats

[Neil Booth <neil at daikokuya dot demon dot co dot uk>]

Joseph S. Myers wrote:-

> Does C++ include hexadecimal floating point constants?  How does it define 
> preprocessing numbers?

I'm 90% sure it's the same as C89.  In other words, no.

> In the course of updating extend.texi to allow for features now being in
> C99, I noticed that G++ objects to hexadecimal floats with -pedantic,
> but still lexes them as numbers in cases where they can't be lexed
> as numbers in C90 (see gcc.dg/c90-hexfloat-2.c), even with -ansi.

Probably set_lang() in cppinit.c needs to be updated with a judicious

      CPP_OPTION (pfile, extended_numbers) = 0;

in the right place.  Then the C++ front end would see 2 tokens instead
of one, so there'd be no warning but a syntax error instead.  I assume
this is what happens for C89.

Neil.