Forgot to install. Now it's okay. Thanks!
Brgds,
Viktor
On 2008.10.27., at 14:52, Przemyslaw Czerpak wrote:
On Mon, 27 Oct 2008, Szak�ts Viktor wrote:
Hi Viktor,
Does it happen with -gc3 only or also with -gc2?
No, only with -gc3. I've only tested this on Darwin PPC,
so I don't know about Darwin Intel, but for sure it
does work for Windows/Intel.
It looks like wrongly set double conversions.
In gencc.c[1096] is:
hb_compGenCString( cargo->yyc, &pFunc->pCode[ lPCodePos + 1 ],
sizeof(
double ) );
It's wrong. Change it to:
{
double d = HB_PCODE_MKDOUBLE( &pFunc->pCode[ lPCodePos + 1 ] );
hb_compGenCString( cargo->yyc, &d, sizeof( double ) );
}
It should help.
I've tried this, but it didn't help for either
problems.
Have you made full clean build?
It should help if -gc2 works correctly.
Maybe the problem isn't with the generated
code, as Darwin/PPC generates exactly the same C
code as the Windows/Intel Harbour.
What is wrong because it does not use IEEE758 little endian double
values.
Yet, the Darwin is wrong when run.
No, the problem is with generated code.
You can also try to recompile Harbour with HB_FORCE_IEEE754_DOUBLE
macro but for me the above modification should be enough.
Please check if it changes generated code. BTW it's a real fix
and should be committed even if it's not enough in this case
what looks strange for me.
best regards,
Przemek
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
