* Fixes
      o Complex number comparisons with NAN operands
      o Cleaned up Phobos installation.
      o Non-virtual method calls
      o Code generation with -mpowerpc64
      o Break in labeled switch statement
* Improvements
      o Updated to DMD 0.110
      o Applied Thomas Kühne's and Anders Björklund's HTML patches.
      o Added Thomas Kühne's "dump source" code
      o Phobos Makefile now supports the DESTDIR variable

In article <crs99u$2b1m$2@digitaldaemon.com>, David Friedman says...
>
>
>* Fixes
>       o Complex number comparisons with NAN operands
>       o Cleaned up Phobos installation.
>       o Non-virtual method calls
>       o Code generation with -mpowerpc64
>       o Break in labeled switch statement
>* Improvements
>       o Updated to DMD 0.110
>       o Applied Thomas Kühne's and Anders Björklund's HTML patches.
>       o Added Thomas Kühne's "dump source" code
>       o Phobos Makefile now supports the DESTDIR variable

Just curiouse (I know here it is), Why don't you post the link to the GDC home page with the announcement?

Ant

Test results for GDC 0.10 on Mac(thanks to Anders) and Linux are online.

http://dstress.kuehne.cn/www/dstress.html

Mac: PASS; Linux: FAIL http://dstress.kuehne.cn/run/opCast_03.d http://dstress.kuehne.cn/run/debug_02.d http://dstress.kuehne.cn/run/command_line_debug_02.d http://dstress.kuehne.cn/run/command_line_debug_04.d

Mac: XFAIL; Linux: XPASS
http://dstress.kuehne.cn/complex/command_line/ [command_line_null_source_04]

Thomas

Ant wrote:
> In article <crs99u$2b1m$2@digitaldaemon.com>, David Friedman says...
> 
>>
>>* Fixes
>>      o Complex number comparisons with NAN operands
>>      o Cleaned up Phobos installation.
>>      o Non-virtual method calls
>>      o Code generation with -mpowerpc64
>>      o Break in labeled switch statement
>>* Improvements
>>      o Updated to DMD 0.110
>>      o Applied Thomas Kühne's and Anders Björklund's HTML patches.
>>      o Added Thomas Kühne's "dump source" code
>>      o Phobos Makefile now supports the DESTDIR variable
> 
> 
> Just curiouse (I know here it is), Why don't you post the
> link to the GDC home page with the announcement?
> 
> Ant
> 
> 

I have been forgetting to post the link.  Thanks for pointing it out!

David

David Friedman wrote:
> 
> * Fixes
>       o Code generation with -mpowerpc64

It built fine with -mcpu=G5 now, but failed
to build when I tried using -mpowerpc64 too:

> expr.c: In function `expand_expr_real':
> expr.c:9128: internal compiler error: in build_mask64_2_operands, at config/rs6000/rs6000.c:2954
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://developer.apple.com/bugreporter> for instructions.

I'm not sure if GCC 3.4.3 is *supposed* to build
on Mac OS X 10.3 with a G5, using that PPC64 flag ?

I know Apple uses 3.3.x themselves, for Panther:
http://developer.apple.com/releasenotes/DeveloperTools/GCC3.html

--anders

I kind of dislike the noise generated during GDC's compilation.

The attached patch eradicates a bunch of pedantic warnings.
(DStress didn't report any regressions)

Thomas

Anders F Björklund wrote:
> David Friedman wrote:
> 
>>
>> * Fixes
>>       o Code generation with -mpowerpc64
> 
> 
> It built fine with -mcpu=G5 now, but failed
> to build when I tried using -mpowerpc64 too:
> 
>> expr.c: In function `expand_expr_real':
>> expr.c:9128: internal compiler error: in build_mask64_2_operands, at config/rs6000/rs6000.c:2954
>> Please submit a full bug report,
>> with preprocessed source if appropriate.
>> See <URL:http://developer.apple.com/bugreporter> for instructions.
> 
> 
> I'm not sure if GCC 3.4.3 is *supposed* to build
> on Mac OS X 10.3 with a G5, using that PPC64 flag ?
> 
> I know Apple uses 3.3.x themselves, for Panther:
> http://developer.apple.com/releasenotes/DeveloperTools/GCC3.html
> 
> --anders

I got the same error (in loop-unroll.c).  This happened with Apple's GCC 3.3.  Using gcc 3.4.3, I was able to do a successful build. However, C++ exceptions do not work when libstdc++ is compiled with -mpowerpc64.  The standard gcc 3.3.3 does not even get past the configure step with -mpowerpc64.

Are you going to submit a bug report to Apple?

David

David Friedman wrote:

>>> expr.c: In function `expand_expr_real':
>>> expr.c:9128: internal compiler error: in build_mask64_2_operands, at config/rs6000/rs6000.c:2954
>>> Please submit a full bug report,
>>> with preprocessed source if appropriate.
>>> See <URL:http://developer.apple.com/bugreporter> for instructions.
>>
> I got the same error (in loop-unroll.c).  This happened with Apple's GCC 3.3.  Using gcc 3.4.3, I was able to do a successful build. However, C++ exceptions do not work when libstdc++ is compiled with -mpowerpc64.  The standard gcc 3.3.3 does not even get past the configure step with -mpowerpc64.

OK - so you need a GCC 3.4 installed, to compile GCC with -mpowerpc64

I didn't try that, as I was content with just using G5 instructions...

> Are you going to submit a bug report to Apple?

I can do that, yes. Or perhaps Darwin, since Apple bugs are secret ?

http://bugzilla.opendarwin.org/

--anders

Hi David, how about submitting your site to www.dmoz.org and yahoo's directory?