IS there a way to see what package versions are available with conda? I am getting an error with jupyter but it was working before. Something like yolk?
To list packages that are installed on your anaconda machine
conda list
This is to list all packages available for anaconda
conda search
You can just type "conda search" which will give you something like the following.
$ conda search
Fetching package metadata .........
affine 2.0.0 py27_0 defaults
2.0.0 py35_0 defaults
2.0.0 py36_0 defaults
alabaster 0.7.3 py27_0 defaults
0.7.3 py34_0 defaults
0.7.7 py27_0 defaults
0.7.7 py34_0 defaults
0.7.7 py35_0 defaults
0.7.9 py27_0 defaults
To trim down the long and slowly loading conda search
output to just the (latest) version(s) appropriate for your environment, you can use MatchSpec filters, as documented here in conda Github repo
For example:
$ conda search "conda-forge::*[name=scikit-learn, subdir=linux-64, build=*py37*]" | tail -n5
scikit-learn 0.21.2 py37h627018c_0 conda-forge
scikit-learn 0.21.2 py37hcdab131_1 conda-forge
scikit-learn 0.21.3 py37hcdab131_0 conda-forge
scikit-learn 0.22 py37hcdab131_0 conda-forge
scikit-learn 0.22 py37hcdab131_1 conda-forge
Note that the most recent version is placed at the bottom of the list (they are sorted in ascending chronological order), so it can be found using tail -n1
, e.g.:
$ conda search "conda-forge::*[name=scikit-learn, subdir=linux-64, build=*py38*]" | tail -n1 | awk {'print $2'}
$ 0.23.2
Cautions:
using version
for narrowing down major and/or minor version is risky, because version=1.*.*
would miss versions such as 1.1
or 1
,
setting architecture (using subdir
key) to linux-64
can miss some useful linux 64-bit packages, if they are stored in the noarch
folder instead of linux-64
To search for a specific package, use: conda search -f <package_name>
. For example, based on the question, to search all versions for "jupyter" package, you'll do: conda search -f jupyter
. This will only return information about packages named "jupyter" exactly.
Source: https://docs.conda.io/projects/conda/en/latest/commands/search.html
If you know the name of the package you want to install search for all available versions of it. eg. for package pandas you will do the following
conda search pandas
and then install the version you want using
conda install pandas=1.0.2
As an addendum, you can use the output of conda search
to fine-tune the version of the package you need installed. E.g. in the list from the 'nasica88', there are three albaster 0.7.7 versions available with with different python versions. If you require e.g. albaster 0.7.7 with python 3.4, you install it as following:
$> conda install albaster=0.7.7=py34_0
So, the second =
sign is your friend here.