07.12.2010 11:58, Jonathan M Davis пишет:
> Okay. I'm trying to get some C code to be properly callable from some D code,
> which naturally involves some extern(C) blocks. One of the types that I have to
> deal with looks something like this:
>
>      typedef struct
>      {
>          unsigned i:1;
>      } my_struct;
>
> I have no idea how to do such a bitfield in D. Does a std.bitmanip.bitfield work?
> I wouldn't have thought so, but I don't know. What would be the proper way to
> create a properly compatible struct in D?
>
> - Jonathan M Davis
>
I'm under the impression that

struct my_struct
{
mixin(bitfields!(
uint, "i", 1,
uint, "", 31));
}

should do the trick.

I haven't tried but maybe htod can help, try copypasting the code to a header file and run it through htod. There's also some flags htod can take to simplify the output, iirc. But maybe htod isn't clever enough to do bitfields..

On 12/7/10, Stanislav Blinov <blinov@loniir.ru> wrote:
> 07.12.2010 11:58, Jonathan M Davis пишет:
>> Okay. I'm trying to get some C code to be properly callable from some D
>> code,
>> which naturally involves some extern(C) blocks. One of the types that I
>> have to
>> deal with looks something like this:
>>
>>      typedef struct
>>      {
>>          unsigned i:1;
>>      } my_struct;
>>
>> I have no idea how to do such a bitfield in D. Does a
>> std.bitmanip.bitfield work?
>> I wouldn't have thought so, but I don't know. What would be the proper way
>> to
>> create a properly compatible struct in D?
>>
>> - Jonathan M Davis
>>
> I'm under the impression that
>
> struct my_struct
> {
> mixin(bitfields!(
> uint, "i", 1,
> uint, "", 31));
> }
>
> should do the trick.
>

Stanislav Blinov:

> I'm under the impression that
> 
> struct my_struct
> {
> mixin(bitfields!(
> uint, "i", 1,
> uint, "", 31));
> }
> 
> should do the trick.

But bitfields in C may not use the same alignments used by bitfields!() on DMD so I think you have to test the sanity of the whole thing for each combination of D compiler, operating system and C compiler.

Bye,
bearophile

bearophile wrote:
> Stanislav Blinov:
> 
>> I'm under the impression that
>>
>> struct my_struct
>> {
>> mixin(bitfields!(
>> uint, "i", 1,
>> uint, "", 31));
>> }
>>
>> should do the trick.
> 
> But bitfields in C may not use the same alignments used by bitfields!() on DMD so I think you have to test the sanity of the whole thing for each combination of D compiler, operating system and C compiler.
> 
> Bye,
> bearophile

hm... what alignments are you talking about? bitfields template uses the  smallest suitable type for all fields, which is uint for the above struct. The trouble may come from bit order, mayhaps. bitfields positions fields in LSb order.

Stanislav Blinov wrote:

>> But bitfields in C may not use the same alignments used by
>> bitfields!() on DMD so I think you have to test the sanity of the
>> whole thing for each combination of D compiler, operating system and C
>> compiler.
>
> hm... what alignments are you talking about? bitfields template uses the
>  smallest suitable type for all fields, which is uint for the above
> struct. The trouble may come from bit order, mayhaps. bitfields
> positions fields in LSb order.

There is almost no guarantee in C (and C++) on how the bitfields are layed out.

I had implemented a C++ template solution once just to overcome that lack of layout. We needed to have our bits layed out exactly in certain ways.

Ali

Andrej Mitrovic wrote:
> I haven't tried but maybe htod can help, try copypasting the code to a
> header file and run it through htod. There's also some flags htod can
> take to simplify the output, iirc. But maybe htod isn't clever enough
> to do bitfields..

htod does bitfields. Give it a try!

On 12/8/10, Walter Bright <newshound2@digitalmars.com> wrote:
> Andrej Mitrovic wrote:
>> I haven't tried but maybe htod can help, try copypasting the code to a header file and run it through htod. There's also some flags htod can take to simplify the output, iirc. But maybe htod isn't clever enough to do bitfields..
>
> htod does bitfields. Give it a try!
>

Cool! But, when is "ctod.exe" coming out? :p

On 12/8/10, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> On 12/8/10, Walter Bright <newshound2@digitalmars.com> wrote:
>> Andrej Mitrovic wrote:
>>> I haven't tried but maybe htod can help, try copypasting the code to a header file and run it through htod. There's also some flags htod can take to simplify the output, iirc. But maybe htod isn't clever enough to do bitfields..
>>
>> htod does bitfields. Give it a try!
>>
>
> Cool! But, when is "ctod.exe" coming out? :p
>

That should have been "cpptod". ;)

Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:

>> Cool! But, when is "ctod.exe" coming out? :p
>>
>
> That should have been "cpptod". ;)

As soon as you write it.

-- 
Simen

On 08/12/2010 02:23, Simen kjaeraas wrote:
> Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>
>>> Cool! But, when is "ctod.exe" coming out? :p
>>>
>>
>> That should have been "cpptod". ;)
>
> As soon as you write it.
>
See d.announce SWIG for D :)