CODEWITHSUNDEEP
pythonconda
nitesh kansal
nitesh kansal

Reputation: 131

'conda install into' downgrades packages why?

I am trying to install 'into' package by using conda install into in the way mentioned in the link. Conda asks me to downgrade the following packages

The following packages will be DOWNGRADED:

    astropy:      1.1.2-np110py27_0  --> 1.0.4-np19py27_0 
    h5py:         2.6.0-np110py27_1  --> 2.5.0-np19py27_3 
    hdf5:         1.8.16-0           --> 1.8.15.1-2       
    libgfortran:  3.0.0-1            --> 3.0-0            
    matplotlib:   1.5.1-np110py27_0  --> 1.4.3-np19py27_2 
    mkl:          11.3.3-0           --> 11.3.1-0         
    mkl-service:  1.1.2-py27_1       --> 1.1.2-py27_0     
    numexpr:      2.5.2-np110py27_1  --> 2.4.4-np19py27_0 
    numpy:        1.10.4-py27_2      --> 1.9.3-py27_2     
    pandas:       0.18.1-np110py27_0 --> 0.17.1-np19py27_0
    pyparsing:    2.1.1-py27_0       --> 2.0.3-py27_0     
    pytables:     3.2.2-np110py27_3  --> 3.2.2-np19py27_0 
    scikit-image: 0.12.3-np110py27_0 --> 0.11.3-np19py27_0
    scikit-learn: 0.17.1-np110py27_1 --> 0.16.1-np19py27_0
    scipy:        0.17.1-np110py27_0 --> 0.17.0-np19py27_2

Can anyone tell me if this is a normal thing? If yes then are the downgraded versions of these packages able to use intel's mkl library?

Upvotes: 1

Views: 285

Answers (1)

Grr
Grr

Reputation: 16079

It looks like almost everything is being altered based on numpy version 1.9 (np19py27_x). Since everything on the left is using numpy 1.10 it looks like the downgrades are in response to a requirement by into to use numpy 1.9. As far as whether the downgrades would work with the mkl library, you would have to check documentation or test it. I would recommend scouring the documentation for mkl library and creating a virtual environment for using into to ensure you don't foul up your default conda setup.

Upvotes: 1

Related Questions