Packages listed multiple times for an environment in anaconda's “conda list”

Question

When I list packages in a conda environment, I get eg the following result:

$ conda list
# packages in environment at /Users/tkb/anaconda3/envs/xxx:
#
_nb_ext_conf              0.4.0                    py35_1  
anaconda-client           1.6.3                    py35_0  
appnope                   0.1.0                    py35_0  
backports.weakref         1.0rc1                    <pip>
bleach                    1.5.0                     <pip>
bleach                    1.5.0                    py35_0  
clyent                    1.2.2                    py35_0  
decorator                 4.0.11                   py35_0  
entrypoints               0.2.3                    py35_0  
html5lib                  0.999                    py35_0  
html5lib                  0.9999999                 <pip>
ipykernel                 4.6.1                    py35_0 
....

Some of the packages are present multiple times with different versions and differ in the third column. What effect does this have, and what does the third column mean?

Answer 1

The third column tells you how the package was installed. If it has pyXX_Y , it was installed by conda, and the XX indicate the Python major and minor version (ie, 35 is Python 3.5), and the build number of the package, which is an internal number used by conda to help judge which versions of a package are newer when the version of the software in the package is the same. If the third column says <pip> , the package was installed by pip. Sometimes, there's a fourth column when the package was installed by conda; this column indicates the channel the package was installed from (specified by -c or --channel in the install or create command).

As for what effect this has, it probably depends on the package. In general, the advice is to install packages into conda environments using conda whenever possible, because conda can't manage (uninstall, upgrade, etc.) packages that were installed by pip.