Shawn Walker wrote:
> OK.  I may see if I can just get it to compile by changing the few 
> 'underscore' errors that show up.  But I will seriously consider the 
> virtualbox.  One thing that bothers me is having to limit the RAM 
> accessible by the virtualbox (this would impact dolfin if I were 
> computing something big).
> 
> Actually, could I just run a virtual terminal on my windows PC that 
> connects to some other linux box?  Orginally, I wanted to avoid this 
> because I was unix retarded.  However, in the process of mucking with 
> all this compiling, I am now less retarded.  Is this a viable option?
>

Yes.

Garth

> - Shawn
> 
> On Wed, 6 Aug 2008, Garth N. Wells wrote:
> 
>> Shawn Walker wrote:
>>> I understand.  I got the umfpack part working.  I was just messing up 
>>> the include directories.  But I still have that other error.
>>>
>>> It has to do with the following code:
>>>
>>> ublas_sparse_matrix& _B = B.mat();
>>>
>>> If I change '_B' to say Bq, then it compiles fine.  For some reason 
>>> it doesn't like the underscore.  I guess this has to do with your 
>>> 'templates' comment. 
>>
>> If you can get it all to work, create a hg bundle and mail it to this 
>> list. Instructions on how to create a bundle can be found at the end 
>> of the DOLFIN manual.
>>
>> If this is just a compiler issue, then I think I
>>> can upgrade the compiler manually to gcc 4.3.
>>>
>>
>> Apparently there are some serious technical obstacles with gcc 4.x and 
>> Cywgin, so I don't expect that this will be trivial.
>>
>>> For this virtual machine, is it a pain in the ass to setup?  It 
>>> looked like it was just an emulator for any operating system you 
>>> want.  Does that mean it will be slow?  I assume it is just as 
>>> flexible as cygwin, if not more.
>>
>> VirtualBox (http://www.virtualbox.org) is free and very easy to use. 
>> My experience is that if you have a relatively new processor and 
>> enable the virtualisation option in the BIOS, performance is near native.
>>
>> Garth
>>
>>>
>>> - Shawn
>>>
>>> On Wed, 6 Aug 2008, Garth N. Wells wrote:
>>>
>>>> Shawn Walker wrote:
>>>>> OK, here we go again.  I disabled UMFPACK, and the compilation goes 
>>>>> further, but I get the following error:
>>>>>
>>>>> ----------------------------------------------
>>>>> $ scons withBoostDir=$BOOST_DIR withUmfpackDir=$UMFPACK_DIR 
>>>>> enableDemos=yes enableScotch=no enableTrilinos=no enableMpi=no 
>>>>> enablePetsc=no enableSlepc=no enableUmfpack=no enableDebug=no
>>>>>
>>>>> scons: Reading SConscript files ...
>>>>> Using options from scons/options.cache
>>>>> Checking for pkg-config... yes
>>>>> Checking for numpy-1... yes
>>>>> Checking for ufc-1... yes
>>>>> Checking for python-2... yes
>>>>> Checking for gts... yes
>>>>> Checking for boost... yes
>>>>> Checking for libxml-2.0... yes
>>>>> Resolving compiler... done
>>>>> Found optional package: gts
>>>>> Unable to find optional package: petsc
>>>>> Unable to find optional package: slepc
>>>>> Unable to find optional package: scotch
>>>>> Unable to find optional package: umfpack
>>>>> Unable to find optional package: trilinos
>>>>> Enabling compilation of PyDOLFIN
>>>>>
>>>>> scons: warning: The env.Copy() method is deprecated; use the 
>>>>> env.Clone() method instead.
>>>>> File "/download/src/fenics/dolfin/dolfin/SConscript", line 107, in 
>>>>> <module>
>>>>> scons: done reading SConscript files.
>>>>> scons: Building targets ...
>>>>> c++ -o dolfin/common/TimeDependent.os -c -Wall -pipe -DNDEBUG -O2 
>>>>> -DPACKAGE_VERSION=\"0.8.0\" -DHAS_GTS=1 -I. 
>>>>> -I/download/src/boost_1_35_0 -I/usr/include/libxml2 
>>>>> -I/build/include -I/usr/include/glib-2.0 
>>>>> -I/usr/lib/glib-2.0/include dolfin/common/TimeDependent.cpp
>>>>>
>>>>> .....................etc...................
>>>>>
>>>>> c++ -o dolfin/ode/MonoAdaptiveJacobian.os -c -Wall -pipe -DNDEBUG 
>>>>> -O2 -DPACKAGE_VERSION=\"0.8.0\" -DHAS_GTS=1 -I. 
>>>>> -I/download/src/boost_1_35_0 -I/usr/include/libxml2 
>>>>> -I/build/include -I/usr/include/glib-2.0 
>>>>> -I/usr/lib/glib-2.0/include dolfin/ode/MonoAdaptiveJacobian.cpp
>>>>> c++ -o dolfin/ode/MonoAdaptiveNewtonSolver.os -c -Wall -pipe 
>>>>> -DNDEBUG -O2 -DPACKAGE_VERSION=\"0.8.0\" -DHAS_GTS=1 -I. 
>>>>> -I/download/src/boost_1_35_0 -I/usr/include/libxml2 
>>>>> -I/build/include -I/usr/include/glib-2.0 
>>>>> -I/usr/lib/glib-2.0/include dolfin/ode/MonoAdaptiveNewtonSolver.cpp
>>>>> dolfin/ode/MonoAdaptiveNewtonSolver.cpp: In member function `void 
>>>>> dolfin::MonoAdaptiveNewtonSolver::debug()':
>>>>> dolfin/ode/MonoAdaptiveNewtonSolver.cpp:265: error: expected 
>>>>> primary-expression before '&' token
>>>>> dolfin/ode/MonoAdaptiveNewtonSolver.cpp:286: error: `128' cannot be 
>>>>> used as a function
>>>>> scons: *** [dolfin/ode/MonoAdaptiveNewtonSolver.os] Error 1
>>>>> scons: building terminated because of errors.
>>>>> ---------------------------------------------------------
>>>>
>>>> I tried this, and received the same error. It looks like like a 
>>>> Cygwin issue with templates, and is perhaps not surprising since 
>>>> Cywgin uses quite old compilers.
>>>>
>>>> I'm not going to spend time on work-arounds for Cygwin peculiarities 
>>>> and for gcc 3.x since running a virtual machine is a viable 
>>>> alternative now and there has been some work on compiling DOLFIN 
>>>> natively under Windows. If someone provides simple patches for 
>>>> Cywgin, we can apply them.
>>>>
>>>> Garth
>>>>
>>>>
>>>>> If there were no errors, run
>>>>>
>>>>>      scons install
>>>>>
>>>>> .....................etc...................
>>>>> ---------------------------------------------------------------
>>>>>
>>>>> Any ideas about this?  Again, I am using Cygwin and the latest 
>>>>> codes through Mercurial.
>>>>>
>>>>> - Shawn
>>>>> _______________________________________________
>>>>> DOLFIN-dev mailing list
>>>>> [email protected]
>>>>> http://www.fenics.org/mailman/listinfo/dolfin-dev
>>>>
>>>>
>>>>
>>


_______________________________________________
DOLFIN-dev mailing list
[email protected]
http://www.fenics.org/mailman/listinfo/dolfin-dev

Reply via email to