It should be noted that only differences are listed in the output. I can verify that the compare command works in conda version 4.8.4. The only mention of this I've found so far is in a pull request that was merged back on July 16, 2020. "Running pkg" is, of course, the one I activated.Įven less clear, the "not found" refers to the active environment: the packages "black", "fribidi", etc, are in my tfm1 (specified) but not in my project/env (active) environment.Ī recent development now provides the ability to compare a conda environment against a requirements file. The documentation is not too clear on this, but "Specification pkg" refers to a package found in the environment whose specification we exported to a file ("tfm1" in my example). Specification pkg: flit-core=3.6.0=pyhd8ed1ab_0, Running pkg: flit-core=3.7.1=pyhd8ed1ab_0įonttools found but mismatch. Specification pkg: executing=0.8.2=pyhd8ed1ab_0, Running pkg: executing=0.8.3=pyhd8ed1ab_0įlit-core found but mismatch.
Next I switch to the other env, in the directory: ❯ conda activate /Users/me/path/project/envĪnd finally use the new compare command: ❯ conda compare /Users/me/path/tfm1_env.ymlĬertifi found but mismatch. ❯ conda env export -f /Users/me/path/tfm1_env.yml
path/project/env with the named env tfm1 (my tensorflow env).įirst I switch to the tfm1 env and export its package list: ❯ conda activate tfm1 Now I want to compare the conda env in the directory. ❯ conda -versionįirst list the environments I have ❯ conda env listĮxperiments /Users/me/miniforge3/envs/experiments It covers the important aspect of using conda and pip together.Using mini conda 4.11, with the new compare command, I'm going to compare a named environment with a path environment. Here is a very good detailed guide that explains how and why to use conda and pip for virtual environments. However, conda will still not index and centrally manage the venv environments, like it does for its own conda environments. You must use "pip install" INSIDE this new conda environment, so conda will index and track those pip package installations. By installing conda into the conda environment, conda will track packages installed by pip into that environment. So now the 'core' conda environment functions like an administrative environment shell. Here I created the conda environment named "core" and installed Python 3.9, conda, and pip into it. You can use conda to create and manage venv and virtualenv environments and other packages installed using pip.įirst create a conda environment with CONDA AND PIP installed into it, e.g., conda create -name core -channel conda-forge python=3.9 conda pip
0 kommentar(er)
