.condarc

# See https://www.anaconda.com/understanding-and-improving-condas-performance/ for more info.

# help debug channel issues
show_channel_urls: true

# pip will always be installed with python 
add_pip_as_python_dependency: true

# strict priority and conda-forge at the top will ensure
# that all of your packages will be from conda-forge unless they only exist on defaults
channel_priority: strict
channels:
  - conda-forge

# when using "conda create" for envs these packages will always be installed
# adjust that list according your needs, the packages below are just a suggestion!
create_default_packages:
  - ipykernel
  - jupyter
  - pip

safety_checks: disabled
auto_activate_base: false

create_condarc.sh

# Add conda-forge and activate strict
conda config --add channels conda-forge
conda config --set channel_priority strict
conda config --set show_channel_urls True

# Make conda faster ny disabeling checks.
conda config --set safety_checks disabled

# Do not actiavte base.
conda config --set auto_activate_base False

# Install these packages by default when using `conda create`.
conda config --append create_default_packages ipykernel \
             --append create_default_packages jupyter \
             --append create_default_packages jupyter_contrib_nbextensions \
             --append create_default_packages pip \
             --append create_default_packages "blas=*=openblas"  # help with the mkl vs openblas issue

I get the following error when running conda update -all after setting up your .condarc

CustomValidationError: Parameter channel_priority = 'strict' declared in <<merged>> is invalid.
The value 'strict' cannot be boolified.

Did I miss something, @ocefpaf?

Did I miss something, @ocefpaf?

What is your conda version? Mine is 4.6.3. I believe that was introduced in 4.6.2 or 4.6.1.

Aha!! 4.5.12... most problems in the world can be solved by prepending conda update conda to them!!! ;-)
Thanks!

Aha!! 4.5.12... most problems in the world can be solved by prepending conda update conda to them!!! ;-)
Thanks!

And create new ones sometimes 😄

But yes, conda evolves fast and using latest is usually desirable.

blas=*=openblas

This is fantastic.

What is the "=*=" operator?

What is the "=*=" operator?

No idea to be honest. I guess that is it whatever needs blas use openblas.
@msarahan can probably explain it better.

I think it means "any version with the openblas build string", i.e., it could be =1.0=openblas, =1.1=openblas, etc.

I think it means "any version with the openblas build string", i.e., it could be =1.0=openblas, =1.1=openblas, etc.

Thanks for the explanation @tadeu!