druntime commit, revision 371


user: andrei

msg:
64-bit compatibility work

http://www.dsource.org/projects/druntime/changeset/371

What is the purpose of sizediff_t, as opposed to ptrdiff_t, which is already defined? Also, Linux uses ssize_t (for "signed" size_t).

dsource.org wrote:
> druntime commit, revision 371
>
>
> user: andrei
>
> msg:
> 64-bit compatibility work
>
> http://www.dsource.org/projects/druntime/changeset/371
>
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime
>
>
> 

Are those math routines all really pure?  I thought they might set floating-point state flags.

On Aug 22, 2010, at 5:13 PM, dsource.org wrote:

> druntime commit, revision 371
> 
> 
> user: andrei
> 
> msg:
> 64-bit compatibility work
> 
> http://www.dsource.org/projects/druntime/changeset/371
> 
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime

In any case, the folloing functions (plus *f/*l variants) can't be pure:

double  frexp(double value, int* exp);
double  modf(double value, double* iptr);
double  remquo(double x, double y, int* quo);
double  nan(char* tagp);

The pure attributes need to be removed from at least these functions. It breaks building druntime on FreeBSD!

src/core/stdc/math.d(494): Error: function core.stdc.math.frexp
(double value, int* exp) is not callable using argument types
(const (real),const (int*))
...etc...


> Are those math routines all really pure?  I thought they might set floating-point state flags.
> 
> On Aug 22, 2010, at 5:13 PM, dsource.org wrote:
> 
> > druntime commit, revision 371
> > 
> > 
> > user: andrei
> > 
> > msg:
> > 64-bit compatibility work
> > 
> > http://www.dsource.org/projects/druntime/changeset/371
> > 
> > _______________________________________________
> > D-runtime mailing list
> > D-runtime at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/d-runtime
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime

On 23 August 2010 00:38, Sean Kelly <sean at invisibleduck.org> wrote:
> Are those math routines all really pure? ?I thought they might set floating-point state flags.

No, they're not pure. Most C math functions are allowed to set the
global variable 'errno'.
I think it's better to assume that NONE are pure.

>
> On Aug 22, 2010, at 5:13 PM, dsource.org wrote:
>
>> druntime commit, revision 371
>>
>>
>> user: andrei
>>
>> msg:
>> 64-bit compatibility work
>>
>> http://www.dsource.org/projects/druntime/changeset/371
>>
>> _______________________________________________
>> D-runtime mailing list
>> D-runtime at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime
>

Okay that's what I thought.  I'll revert this today.

On Aug 25, 2010, at 11:53 PM, Don Clugston wrote:

> On 23 August 2010 00:38, Sean Kelly <sean at invisibleduck.org> wrote:
>> Are those math routines all really pure?  I thought they might set floating-point state flags.
> 
> No, they're not pure. Most C math functions are allowed to set the
> global variable 'errno'.
> I think it's better to assume that NONE are pure.
> 
>> 
>> On Aug 22, 2010, at 5:13 PM, dsource.org wrote:
>> 
>>> druntime commit, revision 371
>>> 
>>> 
>>> user: andrei
>>> 
>>> msg:
>>> 64-bit compatibility work
>>> 
>>> http://www.dsource.org/projects/druntime/changeset/371
>>> 
>>> _______________________________________________
>>> D-runtime mailing list
>>> D-runtime at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>> 
>> _______________________________________________
>> D-runtime mailing list
>> D-runtime at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>> 
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime