A question about the workaround.... Does it mean that I can use sfepy now without any problems if I do not use multiprocessing in the homogenization coefficients calculation?

On Thursday, July 21, 2016 at 11:00:34 AM UTC+2, Robert Cimrman wrote:

On 07/21/2016 10:47 AM, Nikhil Vaidya wrote:
> the workaround suggested by you solves the problem with the
> homogenization_perfusion test but not with the input_linear_elastic_mM test.

Yes, my bad - you need to add the "'multiprocessing' : False," option into
options in linear_homogenization_up.py file - this is the micro-problem used by
input_linear_elastic_mM.

r.

> On Wednesday, July 20, 2016 at 4:02:18 PM UTC+2, Robert Cimrman wrote:
>>
>> Hi Nikhil,
>>
>> to resolve test_projections and test_input_navier_stokes2d_iga, install
>> igakit
>> [1]. But note that it is not needed when using sfepy for finite element
>> computations. You may want igakit only when playing with the isogeometric
>> analysis computations.
>>
>> The other two errors are related to use of multiprocessing when computing
>> homogenized coefficients - I am not sure why that happens. What version of
>> python/multiprocessing do you use?
>>
>> As a workaround you can try adding
>>
>>      'multiprocessing' : False,
>>
>> into options dicts in examples/homogenization/perfusion_micro.py and
>> examples/homogenization/linear_elastic_mM.py.
>>
>> r.
>>
>> On 07/20/2016 12:01 PM, Nikhil Vaidya wrote:
>>> Hello,
>>>
>>> I downloaded the sfepy source code from git and installed it using the
>>> Visual C++ compiler. After installing, I ran the tests and 4 of them
>>> failed. I then ran those tests in debug mode and the output is attached.
>>> Please tell me how to solve this problem.
>>>
>>> Best regards,
>>> Nikhil
>>>
>>
>>
>