On Mon, 04 Jun 2007, Sisyphus wrote:
> On Tue, 05 Jun 2007, "Jan Dubois" <[EMAIL PROTECTED]> wrote:
>> I just verified that at least the 64-bit Platform SDK compilers still
>> use 8 bytes for "long double" variables. Therefore defining
>> uselongdouble would be pointless.
>
> Does that mean it's impossible to build a 64-bit perl (using the
> 64-bit Platform SDK compiler) where the NV provides greater precision
> than the IV ?
Yes, it will be impossible with the platform compilers unless Microsoft
decides to implement better long double support. I would not hold my
breath though:
http://blogs.msdn.com/ericflee/archive/2004/06/10/152852.aspx
> I see that $Config{longdblsize} reports '10'. Does that open a door to
> the possibility of building a perl where the NV uses 10 bytes ?
Those values are bogus. They were hand-written in the canned
win32/config.xx files in the core Perl distribution, and they are wrong,
at least for VC. I guess I should correct at least the VC and VC64
values.
I think it should be possible to build even a 32-bit Perl using MingW
that will use 10-byte long doubles, but I haven't tried it. You
will still run into problems though, as all the printf() style formatting
will break because MingW uses MSVCRT.DLL as the runtime library.
Maybe you could do this with Cygwin. Or with the Intel compiler. I
don't have time to play with this right now, and I also don't expect us
to move ActivePerl on Windows away from VC6/Platform SDK compilers any
time soon.
Cheers,
-Jan
_______________________________________________
Perl-Win32-Users mailing list
[email protected]
To unsubscribe: http://listserv.ActiveState.com/mailman/mysubs
