I see, thanks for the qualification. I have built with gcc4.4.6 all the way
up to 4.7.x so not sure that is the issue. I think boost 1.5.1 is good for
the later releases, certainly 2.x, but again, we built 1.x versions with
1.39 for quite a few years without any issues.
I'm not sure why you would be seeing this linux as is...
Piotr
On 12 February 2014 15:32, Federico Naum <federi...@al.com.au> wrote:
> Hi Piotr,
>
> I think this is more related to what Nick is pointing out (but I'm on
> Linux not in Windows)
>
> Regarding the lines with NAMESPACE that I put in the post, I mix them,
> but just to stress that I'm seeing the same problem either with version 1.X
> and version 2.X of both ilmbase and pylmbase
>
> On verison 1.X there was no namespaces code around that piece of code and
> I'm still seeing the same issue. (crash in the same line)
> But, I don't know what to try, so I'll give it a go to ilmbase 2.1.0, and
> I'll report back
>
> Thanks
> Fede
>
>
>
> On 13/02/14 05:33, Piotr Stanczyk wrote:
>
> Hey -
>
> How did you build IlmBase? Pre- 2.1.0 releases had a bug when building
> with custom namespaces
>
> Piotr
>
>
>
> On 11 February 2014 23:50, Federico Naum <federi...@al.com.au> wrote:
>
>> Hi
>>
>> Hope this is the forum to ask.
>>
>> I'm building pyilmbase 1.0.0 under llinux Centos 6.2, with ilmbase
>> 1.0.3, gcc4.1.2 and boost 1.44.0 (also tried all combinations of the newer
>> version of pyilmbase 2.0.0/1 against ilmbase 2.0 with newer compiler
>> gcc446, and several flavours boost.
>>
>> In all cases I can build successfully and import *iex *just fine, but
>> importing *imath* does SegFault, it does crashed on line 241 of PyIex.h
>>
>> 240│ const TypeTranslator<IEX_NAMESPACE::BaseExc>::ClassDesc
>> *baseDesc = baseExcTranslator().template
>> findClassDesc<ExcBase>(baseExcTranslator().firstClassDesc());
>> *241├> std::string baseName = baseDesc->typeName();*
>> 242│ std::string baseModule = baseDesc->moduleName();
>>
>> *baseDesc* ends up with a nullPointer (const
>> PyIex::TypeTranslator<Iex_2_0::BaseExc>::ClassDesc *) 0x0
>>
>>
>> As importing *iex* works fine, I changed the code in imathmodule.cpp to
>> kind of copy how the Exception gets registered.
>>
>> <
>> PyIex::registerExc<Imath::NullVecExc,Iex::MathExc>("NullVecExc","imath");
>> <
>> PyIex::registerExc<Imath::NullQuatExc,Iex::MathExc>("NullQuatExc","imath");
>> <
>> PyIex::registerExc<Imath::SingMatrixExc,Iex::MathExc>("SingMatrixExc","imath");
>> <
>> PyIex::registerExc<Imath::ZeroScaleExc,Iex::MathExc>("ZeroScaleExc","imath");
>> <
>> PyIex::registerExc<Imath::IntVecNormalizeExc,Iex::MathExc>("IntVecNormalizeExc","imath");
>> ---
>> >
>> PyIex::registerExc<Imath::NullVecExc,Iex::BaseExc>("NullVecExc","iex");
>> >
>> PyIex::registerExc<Imath::NullQuatExc,Iex::BaseExc>("NullQuatExc","iex");
>> >
>> PyIex::registerExc<Imath::SingMatrixExc,Iex::BaseExc>("SingMatrixExc","iex");
>> >
>> PyIex::registerExc<Imath::ZeroScaleExc,Iex::BaseExc>("ZeroScaleExc","iex");
>> >
>> PyIex::registerExc<Imath::IntVecNormalizeExc,Iex::BaseExc>("IntVecNormalizeExc","iex");
>>
>> This way I can import *imath,* and in python land I can see that the
>> *imath* module contains the registered exceptions and I can
>> raise/try/catch them. but this code change seems fishy since nothing else
>> seems to have reported and the code in the latest version remains the same.
>>
>> I saw someone else did the same change (back in 2012)
>> https://groups.google.com/forum/#!searchin/alembic-discussion/Cannot$20build$20windows/alembic-discussion/OMOvhg0Nv7o/VnJRrByVv10J
>>
>> but I'm pretty sure I am missing something more obvious
>>
>> also it caught my attention that I get this warning on gdb
>>
>> *PyIex::TypeTranslator<Iex_2_0::BaseExc>::findClassDesc<Iex_2_0::MathExc>
>> (this=0x6e6510, cd=0x6eaa00) at
>> /scratch/federicon/git/rezExternalPackages/pyilmbase/2.0.0/build/2/pyilmbase-prefix/src/pyilmbase/PyIex/Py*
>> *IexTypeTranslator.h:262*
>> *(gdb) print cd->typeInfo()*
>> *warning: RTTI symbol not found for class
>> 'PyIex::TypeTranslator<Iex_2_0::BaseExc>::ClassDescT<Iex_2_0::BaseExc>'*
>>
>> could it be related to the realtime type info? I made sure that ilmbase
>> and pyilmbase were compiled wihout the -fno-rtti so it should be on by
>> default
>>
>> Hope someone can help me find the issue
>>
>> Thanks in advance
>> Fede
>>
>>
>>
>> _______________________________________________
>> Openexr-devel mailing list
>> Openexr-devel@nongnu.org
>> https://lists.nongnu.org/mailman/listinfo/openexr-devel
>>
>>
>
>
_______________________________________________
Openexr-devel mailing list
Openexr-devel@nongnu.org
https://lists.nongnu.org/mailman/listinfo/openexr-devel
