Hot: inspect fails for newliy written string variables

[ahms5]

inspect fails for newliy written string variables

    sofa_test = sf.Sofa('GeneralTF')
    sofa_test.add_variable('TestString', 'test', 'string', 'S')
    sofa_test.inspect()

in this case sofa_test.inspect()fails, since 'test' does not have size attribute.
Not sure were to fix this, otherwise I would have opend a PR directly.

and this would work:

    sofa_test = sf.Sofa('GeneralTF')
    sofa_test.add_variable('TestString', 'test', 'string', 'S')
    sf.write_sofa(sofa_test, 'test.sofa')
    sofa_test = sf.read_sofa('test.sofa')
    sofa_test.inspect()

[f-brinkmann]

@ahms5 @mberz I updated the response. Can you check if you agree with the suggested solution.

Thanks for reporting. It works when passing the variable as array/list ['test']

sofa_test = sf.Sofa('GeneralTF')
sofa_test.add_variable('TestString', ['test'], 'string', 'S')
sofa_test.inspect()

Passing it as an array is the correct/explicit way, but we allow passing data with a single dimension (such as Data_SamplingRate or the array of a single string in the example above) as a single value for convenience. So your code

sofa_test.add_variable('TestString', 'test', 'string', 'S')

should work in my opinion. In this case TestString is added to the Sofa object as a string, which causes the error in Sofa.inspect. When calling write_sofa, TestString is converted to a numpy array of strings (required for writing a sofa/NetCDF file) and after read_sofa, TestString is of type numpy.str_ and Sofa.inspect works.

I would suggest to make sure that Sofa.add_variable stores the data as a numpy.str_ to solve the problem. This would be implemented in Sofa.__setattr__ and would fix a second related bug that appears when setting the value of an existing string variable. This

sofa = sf.Sofa('AnnotatedReceiverAudio')
sofa.Response = 'Yes'
sofa.inspect()

causes the same error as reported above because it also calles Sofa.__setattr__ internally.