Hi,

I have recently started moving across longdouble.h, and the use of ldouble() over into GDC's copy of the D frontend from upstream DMD.  I see there's an (attempted?) implementation for GDC already in longdouble.h

https://github.com/D-Programming-Language/dmd/blob/master/src/root/longdouble.h#L16

However, this does not work as GDC's real_t implementation relies on some typedef's in mars.h, and #include mars.h won't work as mars.h depends upon longdouble.h.  The only way I can see around it is to split off all integral and floating typedefs (perhaps not complex_t) from mars.h into a new header named - eg, coretypes.h.  Would it be ok if I perform this slight internal move about and pull it into DMD?


Regards
-- 
Iain Buclaw

*(p < e ? p++ : p) = (c & 0x0f) + '0';
_______________________________________________
dmd-internals mailing list
dmd-internals@puremagic.com
http://lists.puremagic.com/mailman/listinfo/dmd-internals

On 18.07.2012 09:13, Iain Buclaw wrote:
> Hi,
>
> I have recently started moving across longdouble.h, and the use of
> ldouble() over into GDC's copy of the D frontend from upstream DMD.  I
> see there's an (attempted?) implementation for GDC already in
> longdouble.h
>
> https://github.com/D-Programming-Language/dmd/blob/master/src/root/longdouble.h#L16

This was just guesswork, as I could not try it out.

I wonder why the real_t implementation is necessary at all, as GCC seems to be capable of 80-bit floating points with "long double". At least that is how DMD is built on posix systems. So maybe you can also use the !_MSC_VER branch of the conditional.

>
> However, this does not work as GDC's real_t implementation relies on
> some typedef's in mars.h, and #include mars.h won't work as mars.h
> depends upon longdouble.h.  The only way I can see around it is to
> split off all integral and floating typedefs (perhaps not complex_t)
> from mars.h into a new header named - eg, coretypes.h.  Would it be ok
> if I perform this slight internal move about and pull it into DMD?
>
>
> Regards
>

_______________________________________________
dmd-internals mailing list
dmd-internals@puremagic.com
http://lists.puremagic.com/mailman/listinfo/dmd-internals

On 18 July 2012 18:46, Rainer Schuetze <r.sagitario@gmx.de> wrote:
> On 18.07.2012 09:13, Iain Buclaw wrote:
>>
>> Hi,
>>
>> I have recently started moving across longdouble.h, and the use of ldouble() over into GDC's copy of the D frontend from upstream DMD.  I see there's an (attempted?) implementation for GDC already in longdouble.h
>>
>>
>> https://github.com/D-Programming-Language/dmd/blob/master/src/root/longdouble.h#L16
>
>
> This was just guesswork, as I could not try it out.
>
> I wonder why the real_t implementation is necessary at all, as GCC seems to be capable of 80-bit floating points with "long double". At least that is how DMD is built on posix systems. So maybe you can also use the !_MSC_VER branch of the conditional.
>
>

It's there because there are a variety of representations for floating point numbers. This means that in a cross-compiler the representation of floating point numbers in the compiled program may be different from that used in the machine doing the compilation.  Therefore, a cross compiler cannot safely use the host machine's floating point arithmetic; it must emulate the target's arithmetic. To ensure consistency, GCC always uses emulation to work with floating point values, even when the host and target floating point formats are identical.


-- 
Iain Buclaw

*(p < e ? p++ : p) = (c & 0x0f) + '0';
_______________________________________________
dmd-internals mailing list
dmd-internals@puremagic.com
http://lists.puremagic.com/mailman/listinfo/dmd-internals

On 18 July 2012 19:13, Iain Buclaw <ibuclaw@ubuntu.com> wrote:
> On 18 July 2012 18:46, Rainer Schuetze <r.sagitario@gmx.de> wrote:
>> On 18.07.2012 09:13, Iain Buclaw wrote:
>>>
>>> Hi,
>>>
>>> I have recently started moving across longdouble.h, and the use of ldouble() over into GDC's copy of the D frontend from upstream DMD.  I see there's an (attempted?) implementation for GDC already in longdouble.h
>>>
>>>
>>> https://github.com/D-Programming-Language/dmd/blob/master/src/root/longdouble.h#L16
>>
>>
>> This was just guesswork, as I could not try it out.
>>
>> I wonder why the real_t implementation is necessary at all, as GCC seems to be capable of 80-bit floating points with "long double". At least that is how DMD is built on posix systems. So maybe you can also use the !_MSC_VER branch of the conditional.
>>
>>
>
> It's there because there are a variety of representations for floating point numbers. This means that in a cross-compiler the representation of floating point numbers in the compiled program may be different from that used in the machine doing the compilation.  Therefore, a cross compiler cannot safely use the host machine's floating point arithmetic; it must emulate the target's arithmetic. To ensure consistency, GCC always uses emulation to work with floating point values, even when the host and target floating point formats are identical.
>

I've actually found no harm putting it in the gdc real_t header instead. So don't need the longdouble header afterall. :)

https://github.com/D-Programming-GDC/GDC/commit/d3dc166ff6f533712b07e05570cb9995c29b441b#L2R101


-- 
Iain Buclaw

*(p < e ? p++ : p) = (c & 0x0f) + '0';
_______________________________________________
dmd-internals mailing list
dmd-internals@puremagic.com
http://lists.puremagic.com/mailman/listinfo/dmd-internals