Re: Failures in tests for obj-c++....

[David Ayers <d dot ayers at inode dot at>]

David Ayers wrote:

> Ziemowit Laski wrote:
> 
>>On 6 Jun 2005, at 22.26, Christian Joensson wrote:
>>
> 
> [snip]
> 
>>>./bitfield-4.exe: error while loading shared libraries: libobjc.so.1:
>>>cannot open shared object file: No such file or directory
>>>FAIL: obj-c++.dg/bitfield-4.mm execution test
>>>
>>>Any ideas of what migt go wrong?
>>
>>
>>No idea. :-(  Perhaps someone maintaining the GNU runtime could take a 
>>look, and also address the i686-pc-linux-gnu ObjC/ObjC++ failures 
>>(http://gcc.gnu.org/ml/gcc/2005-05/msg01513.html)...
> 
> 
> I cannot reproduce this, yet I think this patch is probably correct
> anyway.  Does it help?

OK, besides the fact that the patch was supposed to go to gcc-patches@,
I think this may not really address the issue.  I could imagine that the
problem lies in using: #include <objc/hash.h> in some header files
instead of #include "hash.h".  (Or issue with the ordering of the -I
directives.)  Could you test whether it's pulling the wrong headers by
poisoning the installed headers and maybe posting the compile invocation
for selector.c, archive.c or init.c?

Thanks,
David Ayers