On Tuesday, 23 September 2014 at 20:29:22 UTC, Walter Bright wrote:
> On 9/23/2014 12:57 PM, Jacob Carlborg wrote:
>> From the LLVM docs [1]:
>
> Thank you. This is helpful.

Yes, that's quite a nice list already.

As far as the C++ implementation details go, I posted the most interesting source files in the earlier discussion: http://forum.dlang.org/thread/lutf3c$2usj$1@digitalmars.com (the GCC ones can be found in the GCC release tarballs, the OS X Clang ones at http://llvm.org/svn/llvm-project/libcxxabi/trunk/src/).

David

On 9/23/14, 2:07 PM, David Nadlinger wrote:
> On Tuesday, 23 September 2014 at 20:29:22 UTC, Walter Bright wrote:
>> On 9/23/2014 12:57 PM, Jacob Carlborg wrote:
>>> From the LLVM docs [1]:
>>
>> Thank you. This is helpful.
>
> Yes, that's quite a nice list already.
>
> As far as the C++ implementation details go, I posted the most
> interesting source files in the earlier discussion:
> http://forum.dlang.org/thread/lutf3c$2usj$1@digitalmars.com (the GCC
> ones can be found in the GCC release tarballs, the OS X Clang ones at
> http://llvm.org/svn/llvm-project/libcxxabi/trunk/src/).
>
> David

Speaking of which, what's the current story with gdc and ldc with regard to unwinding? Can they unwind properly when a C++ exception gets thrown? -- Andrei

On Tuesday, 23 September 2014 at 21:20:37 UTC, Andrei Alexandrescu wrote:
> Speaking of which, what's the current story with gdc and ldc with regard to unwinding? Can they unwind properly when a C++ exception gets thrown? -- Andrei

Currently, the LDC runtime just aborts, but that wouldn't be too hard to fix.

David

On Tuesday, 23 September 2014 at 17:37:42 UTC, Andrei
Alexandrescu wrote:
> We need a libunwind expert to figure out a good approach for handling exceptions thrown by C++ code into D.
>
> Is anyone fluent with libunwind?
>
>
> Andrei

https://github.com/deadalnix/libsdrt/blob/master/src/d/rt/eh.d

On Tuesday, 23 September 2014 at 18:14:35 UTC, Andrei
Alexandrescu wrote:
> On 9/23/14, 11:03 AM, David Nadlinger wrote:
>> On Tuesday, 23 September 2014 at 17:37:42 UTC, Andrei Alexandrescu wrote:
>>> We need a libunwind expert to figure out a good approach for handling
>>> exceptions thrown by C++ code into D.
>>>
>>> Is anyone fluent with libunwind?
>>
>> More or less.
>>
>> What exactly are your goals? I thought the options we have are fairly
>> clear, as discussed a few days ago. At least, Amaury and I seemed to
>> agree (both of us worked on EH recently). The main question to decide is
>> whether you want to go down the catch-C++-exceptions-in-D rabbit hole.
>
> I think we should explore that rabbit hole. In the interim, a simpler solution may work: a C++ exception can only be caught from within C++ code. If no handler, the program terminates reliably. -- Andrei

Here is the other side of the rabbit hole you want to be
compatible with: libstdc++-v3//libsupc++/eh_personality.cc (from
GCC source tree, svn checkout svn://gcc.gnu.org/svn/gcc/trunk
SomeLocalDir ).

I think catching C++ exception may be realistic on the long run,
but having them unwind properly should probably be the goal on
the short run.

On 23 September 2014 22:20, Andrei Alexandrescu via Digitalmars-d <digitalmars-d@puremagic.com> wrote:
> On 9/23/14, 2:07 PM, David Nadlinger wrote:
>>
>> On Tuesday, 23 September 2014 at 20:29:22 UTC, Walter Bright wrote:
>>>
>>> On 9/23/2014 12:57 PM, Jacob Carlborg wrote:
>>>>
>>>> From the LLVM docs [1]:
>>>
>>>
>>> Thank you. This is helpful.
>>
>>
>> Yes, that's quite a nice list already.
>>
>> As far as the C++ implementation details go, I posted the most interesting source files in the earlier discussion: http://forum.dlang.org/thread/lutf3c$2usj$1@digitalmars.com (the GCC ones can be found in the GCC release tarballs, the OS X Clang ones at http://llvm.org/svn/llvm-project/libcxxabi/trunk/src/).
>>
>> David
>
>
> Speaking of which, what's the current story with gdc and ldc with regard to unwinding? Can they unwind properly when a C++ exception gets thrown? -- Andrei

GDC lets foreign exceptions pass through just fine.  The only time when runtime terminates is when GDC has no choice *but* to handle a foreign exception it doesn't know about.


By all means, you can have this kind of code:
---
D:
extern(C++) int Dfunction()
{
  try {
      return CXXfunction(2);
  }
  catch {
      return 8;
  }
}
---
C++:
int main() {
  try {
      return Dfunction();
  }
  catch (int error) {
      return error;
  }
}

int CXXfunction(int i)
{
  throw int(16);
  return i * 2;
}

And the runtime flow is:
Phase 1.
=> __gdc_personality_v0   // Found nothing, continue unwinding.
=> __gxx_personality_v0   // Found something, cache details and continue unwind

Phase 2.
=> __gdc_personality_v0   // Found nothing, continue unwinding.
=> __gxx_personality_v0   // Found something, finish unwinding.

And we land back in C++ main, return 16.

Iain.

On 23 September 2014 22:29, deadalnix via Digitalmars-d <digitalmars-d@puremagic.com> wrote:
> On Tuesday, 23 September 2014 at 17:37:42 UTC, Andrei Alexandrescu wrote:
>>
>> We need a libunwind expert to figure out a good approach for handling exceptions thrown by C++ code into D.
>>
>> Is anyone fluent with libunwind?
>>
>>
>> Andrei
>
>
> https://github.com/deadalnix/libsdrt/blob/master/src/d/rt/eh.d

https://github.com/D-Programming-GDC/GDC/blob/master/libphobos/libdruntime/gcc/deh.d

:o)

On 9/23/14, 3:14 PM, Iain Buclaw via Digitalmars-d wrote:
> GDC lets foreign exceptions pass through just fine.

Proper unwinding of the D stack and all? -- Andrei

On 9/23/14, 3:15 PM, Iain Buclaw via Digitalmars-d wrote:
> On 23 September 2014 22:29, deadalnix via Digitalmars-d
> <digitalmars-d@puremagic.com> wrote:
>> On Tuesday, 23 September 2014 at 17:37:42 UTC, Andrei
>> Alexandrescu wrote:
>>>
>>> We need a libunwind expert to figure out a good approach for handling
>>> exceptions thrown by C++ code into D.
>>>
>>> Is anyone fluent with libunwind?
>>>
>>>
>>> Andrei
>>
>>
>> https://github.com/deadalnix/libsdrt/blob/master/src/d/rt/eh.d
>
> https://github.com/D-Programming-GDC/GDC/blob/master/libphobos/libdruntime/gcc/deh.d
>
> :o)

Noice. How difficult would be to port this to dmd/Linux? This is big - we need to get it done, and soon.

Andrei

On 9/23/2014 3:45 PM, Andrei Alexandrescu wrote:
> On 9/23/14, 3:15 PM, Iain Buclaw via Digitalmars-d wrote:
>> On 23 September 2014 22:29, deadalnix via Digitalmars-d
>> <digitalmars-d@puremagic.com> wrote:
>>> On Tuesday, 23 September 2014 at 17:37:42 UTC, Andrei
>>> Alexandrescu wrote:
>>>>
>>>> We need a libunwind expert to figure out a good approach for handling
>>>> exceptions thrown by C++ code into D.
>>>>
>>>> Is anyone fluent with libunwind?
>>>>
>>>>
>>>> Andrei
>>>
>>>
>>> https://github.com/deadalnix/libsdrt/blob/master/src/d/rt/eh.d
>>
>> https://github.com/D-Programming-GDC/GDC/blob/master/libphobos/libdruntime/gcc/deh.d
>>
>>
>> :o)
>
> Noice. How difficult would be to port this to dmd/Linux? This is big - we need
> to get it done, and soon.
>
> Andrei

There is a problem, that code is gpl licensed. It may not actually matter, since this code is only to interface with gpl libraries, but we should be careful.