mamba/micromamba does not honour "nodefaults" in channel list

[pauldmccarthy]

Troubleshooting docs

• My problem is not solved in the Troubleshooting docs

Anaconda default channels

• I do NOT use the Anaconda default channels (pkgs/* etc.)

How did you install Mamba?

Micromamba

Search tried in issue tracker

nodefaults

Latest version of Mamba

• My problem is not solved with the latest version

Tried in Conda?

I do not have this problem with Conda, just with Mamba

Describe your issue

Howdy,

It seems that micromamba does not honour nodefaults, which we use to disable querying of the anaconda defaults channels. In my specific use case, we are trying to override any user settings which may be in ~/.condarc with a root prefix .condarc that contains nodefaults. So perhaps the issue is in the way that the configuration files are merged.

I have devised a minimal example which exhibits the behaviour, and which is described below. I've been testing in an almalinux:latest docker container (e.g. docker run -it almalinux:latest /bin/bash):

# First install some basic system utilities
dnf update
dnf install --allowerasing -y curl bzip2 tar

# Then download and install a micromamba root envirionment

MMURL=https://micro.mamba.pm/api/micromamba/linux-64/latest
export MAMBA_ROOT_PREFIX=/micromamba
mkdir ${MAMBA_ROOT_PREFIX}
curl -Ls ${MMURL} | tar -C ${MAMBA_ROOT_PREFIX} -xvj bin/micromamba

eval "$(${MAMBA_ROOT_PREFIX}/bin/micromamba shell hook --shell $(basename ${SHELL}))"

# Now set up a user `.condarc` file which lists `defaults`,
# and a root prefix `.condarc` file which lists `nodefaults`.
# I would expect the latter to override the former:

echo "channels:"    > ~/.condarc
echo " - defaults" >> ~/.condarc

echo 'channels:      #!final'   > ${MAMBA_ROOT_PREFIX}/.condarc
echo ' - conda-forge #!top'    >> ${MAMBA_ROOT_PREFIX}/.condarc
echo ' - nodefaults  #!bottom' >> ${MAMBA_ROOT_PREFIX}/.condarc

Now let's try and install some packages - I would expect them to be sourced from conda-forge:

micromamba install -p ${MAMBA_ROOT_PREFIX} python numpy

But no, micromamba proposes to install the packages from main:

Details

warning  libmamba 'repo.anaconda.com', a commercial channel hosted by Anaconda.com, is used.

warning  libmamba Please make sure you understand Anaconda Terms of Services.

warning  libmamba See: https://legal.anaconda.com/policies/en/
warning  libmamba 'repo.anaconda.com', a commercial channel hosted by Anaconda.com, is used.

warning  libmamba Please make sure you understand Anaconda Terms of Services.

warning  libmamba See: https://legal.anaconda.com/policies/en/
warning  libmamba 'repo.anaconda.com', a commercial channel hosted by Anaconda.com, is used.

warning  libmamba Please make sure you understand Anaconda Terms of Services.

warning  libmamba See: https://legal.anaconda.com/policies/en/
warning  libmamba 'repo.anaconda.com', a commercial channel hosted by Anaconda.com, is used.

warning  libmamba Please make sure you understand Anaconda Terms of Services.

warning  libmamba See: https://legal.anaconda.com/policies/en/
pkgs/main/noarch                                   741.7kB @ 668.5kB/s  1.1s
pkgs/r/noarch                                        2.0MB @ 908.4kB/s  2.2s
pkgs/r/linux-64                                      1.6MB @ 664.4kB/s  2.4s
nodefaults/linux-64                                 ??.?MB @  ??.?MB/s  0.8s
nodefaults/noarch                                   ??.?MB @  ??.?MB/s  0.8s
conda-forge/noarch                                  19.2MB @   2.5MB/s  7.6s
pkgs/main/linux-64                                   7.1MB @ 702.0kB/s 10.1s
conda-forge/linux-64                                42.2MB @   2.2MB/s 19.3s


Transaction

  Prefix: /micromamba

  Updating specs:

   - python
   - numpy


  Package                Version  Build               Channel        Size
───────────────────────────────────────────────────────────────────────────
  Install:
───────────────────────────────────────────────────────────────────────────

  + _libgcc_mutex            0.1  main                pkgs/main       3kB
  + _openmp_mutex            5.1  1_gnu               pkgs/main      21kB
  + blas                     1.0  openblas            pkgs/main      47kB
  + bzip2                  1.0.8  h5eee18b_6          pkgs/main     268kB
  + ca-certificates   2024.12.31  h06a4308_0          pkgs/main     131kB
  + expat                  2.6.4  h6a678d5_0          pkgs/main     184kB
  + ld_impl_linux-64        2.40  h12ee557_0          pkgs/main     727kB
  + libffi                 3.4.4  h6a678d5_1          pkgs/main     145kB
  + libgcc-ng             11.2.0  h1234567_1          pkgs/main       6MB
  + libgfortran-ng        11.2.0  h00389a5_1          pkgs/main      20kB
  + libgfortran5          11.2.0  h1234567_1          pkgs/main       2MB
  + libgomp               11.2.0  h1234567_1          pkgs/main     485kB
  + libmpdec               4.0.0  h5eee18b_0          pkgs/main      88kB
  + libopenblas           0.3.21  h043d6bf_0          pkgs/main       6MB
  + libstdcxx-ng          11.2.0  h1234567_1          pkgs/main       5MB
  + libuuid               1.41.5  h5eee18b_0          pkgs/main      28kB
  + ncurses                  6.4  h6a678d5_0          pkgs/main     936kB
  + numpy                  2.2.2  py313h3a69d60_0     pkgs/main      11kB
  + numpy-base             2.2.2  py313habf87d0_0     pkgs/main       9MB
  + openssl               3.0.15  h5eee18b_0          pkgs/main       5MB
  + pip                     25.0  py313h06a4308_0     pkgs/main       3MB
  + python                3.13.2  hf623796_100_cp313  pkgs/main      38MB
  + python_abi              3.13  0_cp313             pkgs/main       6kB
  + readline                 8.2  h5eee18b_0          pkgs/main     365kB
  + setuptools            72.1.0  py313h06a4308_0     pkgs/main       3MB
  + sqlite                3.45.3  h5eee18b_0          pkgs/main       1MB
  + tk                    8.6.14  h39e8969_0          pkgs/main       4MB
  + tzdata                 2025a  h04d1e81_0          pkgs/main     120kB
  + wheel                 0.45.1  py313h06a4308_0     pkgs/main     148kB
  + xz                     5.6.4  h5eee18b_1          pkgs/main     580kB
  + zlib                  1.2.13  h5eee18b_1          pkgs/main     113kB

  Summary:

  Install: 31 packages

  Total download: 85MB

───────────────────────────────────────────────────────────────────────────

Output of micromamba config list:

channels:
  - defaults
  - conda-forge
  - nodefaults

Output of micromamba config sources:

Configuration files (by precedence order):
~/.condarc
/micromamba/.condarc

Output of micromamba info:

Details

       libmamba version : 2.0.5
     micromamba version : 2.0.5
           curl version : libcurl/8.11.1 OpenSSL/3.4.0 zlib/1.3.1 zstd/1.5.6 libssh2/1.11.1 nghttp2/1.64.0
     libarchive version : libarchive 3.7.7 zlib/1.3.1 bz2lib/1.0.8 libzstd/1.5.6
       envs directories : /micromamba/envs
          package cache : /micromamba/pkgs
                          /root/.mamba/pkgs
            environment : base
           env location : /micromamba
      user config files : /root/.mambarc
 populated config files : /root/.condarc
                          /micromamba/.condarc
       virtual packages : __unix=0=0
                          __linux=6.8.0=0
                          __glibc=2.34=0
                          __archspec=1=x86_64_v3
               channels : https://repo.anaconda.com/pkgs/main/linux-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/linux-64
                          https://repo.anaconda.com/pkgs/r/noarch
                          https://conda.anaconda.org/conda-forge/linux-64
                          https://conda.anaconda.org/conda-forge/noarch
                          https://conda.anaconda.org/nodefaults/linux-64
                          https://conda.anaconda.org/nodefaults/noarch
       base environment : /micromamba
               platform : linux-64

I can perform the same experiement with Miniforge instead of micromamba (this is in a new docker container):

dnf update
dnf install --allowerasing -y curl bzip2 tar

curl -OL https://github.com/conda-forge/miniforge/releases/download/25.1.1-0/Miniforge3-25.1.1-0-Linux-x86_64.sh

sh Miniforge3-25.1.1-0-Linux-x86_64.sh -b -p /miniforge3/
echo "channels:"    > ~/.condarc
echo " - defaults" >> ~/.condarc

echo 'channels:      #!final'   > /miniforge3/.condarc
echo ' - conda-forge #!top'    >> /miniforge3/.condarc
echo ' - nodefaults  #!bottom' >> /miniforge3/.condarc

Now attempting to install the packages:

/miniforge3/bin/conda install python numpy

I get the expected behaviour, i.e. the requested packages are sourced from conda-forge:

Details

Channels:
 - conda-forge
 - nodefaults
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: /miniforge3

  added / updated specs:
    - numpy
    - python


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    certifi-2025.1.31          |     pyhd8ed1ab_0         159 KB  conda-forge
    conda-25.1.1               |  py312h7900ff3_1         1.1 MB  conda-forge
    libblas-3.9.0              |31_h59b9bed_openblas          16 KB  conda-forge
    libcblas-3.9.0             |31_he106b2a_openblas          16 KB  conda-forge
    libgfortran-14.2.0         |       h69a702a_2          52 KB  conda-forge
    libgfortran5-14.2.0        |       hf1ad2bd_2         1.4 MB  conda-forge
    liblapack-3.9.0            |31_h7ac8fdf_openblas          16 KB  conda-forge
    libopenblas-0.3.29         |pthreads_h94d23a6_0         5.6 MB  conda-forge
    numpy-2.2.3                |  py312h72c5963_0         8.1 MB  conda-forge
    openssl-3.4.1              |       h7b32b05_0         2.8 MB  conda-forge
    python-3.12.9              |h9e4cc4f_0_cpython        30.1 MB  conda-forge
    ------------------------------------------------------------
                                           Total:        49.4 MB

The following NEW packages will be INSTALLED:

  libblas            conda-forge/linux-64::libblas-3.9.0-31_h59b9bed_openblas
  libcblas           conda-forge/linux-64::libcblas-3.9.0-31_he106b2a_openblas
  libgfortran        conda-forge/linux-64::libgfortran-14.2.0-h69a702a_2
  libgfortran5       conda-forge/linux-64::libgfortran5-14.2.0-hf1ad2bd_2
  liblapack          conda-forge/linux-64::liblapack-3.9.0-31_h7ac8fdf_openblas
  libopenblas        conda-forge/linux-64::libopenblas-0.3.29-pthreads_h94d23a6_0
  numpy              conda-forge/linux-64::numpy-2.2.3-py312h72c5963_0

The following packages will be UPDATED:

  certifi                           2024.12.14-pyhd8ed1ab_0 --> 2025.1.31-pyhd8ed1ab_0
  conda                              25.1.1-py312h7900ff3_0 --> 25.1.1-py312h7900ff3_1
  openssl                                  3.4.0-h7b32b05_1 --> 3.4.1-h7b32b05_0
  python                          3.12.8-h9e4cc4f_1_cpython --> 3.12.9-h9e4cc4f_0_cpython

Output of mamba config list and config sources:

[root@237ea3984070 /]# /miniforge3/bin/mamba config list
channels:
  - defaults
  - conda-forge
  - nodefaults
[root@237ea3984070 /]# /miniforge3/bin/mamba config sources
Configuration files (by precedence order):
~/.condarc
/miniforge3/.condarc

Output of mamba info:

Details

[root@237ea3984070 /]# /miniforge3/bin/mamba info

       libmamba version : 2.0.6
          mamba version : 2.0.6
           curl version : libcurl/8.11.1 OpenSSL/3.4.0 zlib/1.3.1 zstd/1.5.6 libssh2/1.11.1 nghttp2/1.64.0
     libarchive version : libarchive 3.7.7 zlib/1.3.1 liblzma/5.6.3 bz2lib/1.0.8 liblz4/1.10.0 libzstd/1.5.6
       envs directories : /miniforge3/envs
          package cache : /miniforge3/pkgs
                          /root/.mamba/pkgs
            environment : base
           env location : /miniforge3
      user config files : /root/.mambarc
 populated config files : /root/.condarc
                          /miniforge3/.condarc
       virtual packages : __unix=0=0
                          __linux=6.8.0=0
                          __glibc=2.34=0
                          __archspec=1=x86_64_v3
               channels : https://repo.anaconda.com/pkgs/main/linux-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/linux-64
                          https://repo.anaconda.com/pkgs/r/noarch
                          https://conda.anaconda.org/conda-forge/linux-64
                          https://conda.anaconda.org/conda-forge/noarch
                          https://conda.anaconda.org/nodefaults/linux-64
                          https://conda.anaconda.org/nodefaults/noarch
       base environment : /miniforge3
               platform : linux-64

For context, we are using micromamba to install our software on end-user systems; in order to make our software installation immune to any settings which may be present in the user's ~/.condarc file, we install a carefully crafted .condarc file into ${MAMBA_ROOT_PREFIX}, which contains our channel list, and includes nodefaults. We have recently switched over from miniforge/miniconda to micromamba, and in general are much happier - we really appreciate the hard work you've been doing on it!

[pauldmccarthy]

Update: In the second experiment, where I was using miniforge, I accidentally ran /miniforge/bin/conda install numpy python instead of /miniforge/bin/mamba install numpy python. Running conda produced the desired behaviour, i.e. honouring nodefaults. However, it seems that mamba has the same problem as micromamba:

/miniforge3/bin/mamba install -p /miniforge3/ python numpy

Details

warning  libmamba 'repo.anaconda.com', a commercial channel hosted by Anaconda.com, is used.

warning  libmamba Please make sure you understand Anaconda Terms of Services.

warning  libmamba See: https://legal.anaconda.com/policies/en/
pkgs/main/noarch                                   741.7kB @ 654.2kB/s  1.0s
pkgs/r/linux-64                                      1.6MB @ 775.5kB/s  1.9s
pkgs/r/noarch                                        2.0MB @   1.0MB/s  1.9s
nodefaults/noarch                                   ??.?MB @  ??.?MB/s  0.6s
nodefaults/linux-64                                 ??.?MB @  ??.?MB/s  0.6s
pkgs/main/linux-64                                   7.1MB @ 765.7kB/s  9.3s
conda-forge/noarch                                  19.2MB @   2.2MB/s  8.6s
conda-forge/linux-64                                42.2MB @   2.2MB/s 19.3s


Transaction

  Prefix: /miniforge3

  Updating specs:

   - python
   - numpy


  Package                       Version  Build               Channel           Size
─────────────────────────────────────────────────────────────────────────────────────
  Install:
─────────────────────────────────────────────────────────────────────────────────────

  + blas                            1.0  openblas            pkgs/main         47kB
  + expat                         2.6.4  h6a678d5_0          pkgs/main        184kB
  + icu                            73.1  h6a678d5_0          pkgs/main         27MB
  + libgfortran-ng               11.2.0  h00389a5_1          pkgs/main         20kB
  + libgfortran5                 11.2.0  h1234567_1          pkgs/main          2MB
  + libmpdec                      4.0.0  h5eee18b_0          pkgs/main         88kB
  + libopenblas                  0.3.21  h043d6bf_0          pkgs/main          6MB
  + numpy                         2.2.2  py313h3a69d60_0     pkgs/main         11kB
  + numpy-base                    2.2.2  py313habf87d0_0     pkgs/main          9MB
  + pcre2                         10.42  hebb0a14_1          pkgs/main          1MB
  + sqlite                       3.45.3  h5eee18b_0          pkgs/main          1MB
  + xz                            5.6.4  h5eee18b_1          pkgs/main        580kB
  + zlib                         1.2.13  h4ab18f5_6          conda-forge       93kB

  Change:
─────────────────────────────────────────────────────────────────────────────────────

  - cffi                         1.17.1  py312h06ac9bb_0     conda-forge     Cached
  + cffi                         1.17.1  py313h1fdaa30_1     pkgs/main        312kB
  - conda                        25.1.1  py312h7900ff3_0     conda-forge     Cached
  + conda                        25.1.1  py313h78bf25f_1     conda-forge        1MB
  - libarchive                    3.7.7  h4585015_3          conda-forge     Cached
  + libarchive                    3.7.7  hfab0078_0          pkgs/main        959kB
  - libsolv                      0.7.30  h3509ff9_0          conda-forge     Cached
  + libsolv                      0.7.30  he621ea3_1          pkgs/main        503kB
  - libssh2                      1.11.1  hf672d98_0          conda-forge     Cached
  + libssh2                      1.11.1  h251f7ec_0          pkgs/main        315kB
  - libxml2                      2.13.5  h0d44e9d_1          conda-forge     Cached
  + libxml2                      2.13.5  hfdd30dd_0          pkgs/main        756kB
  - menuinst                      2.2.0  py312h7900ff3_0     conda-forge     Cached
  + menuinst                      2.2.0  py313h06a4308_1     pkgs/main        173kB
  - pip                            25.0  pyh8b19718_0        conda-forge     Cached
  + pip                            25.0  py313h06a4308_0     pkgs/main          3MB
  - pycosat                       0.6.6  py312h66e93f0_2     conda-forge     Cached
  + pycosat                       0.6.6  py313h5eee18b_2     pkgs/main         93kB
  - ruamel.yaml.clib              0.2.8  py312h66e93f0_1     conda-forge     Cached
  + ruamel.yaml.clib              0.2.8  py313h536fd9c_1     conda-forge      144kB
  - zstandard                    0.23.0  py312hef9b889_1     conda-forge     Cached
  + zstandard                    0.23.0  py313h2c38b39_1     pkgs/main        455kB

  Reinstall:
─────────────────────────────────────────────────────────────────────────────────────

  o archspec                      0.2.5  pyhd8ed1ab_0        conda-forge     Cached
  o boltons                      24.0.0  pyhd8ed1ab_1        conda-forge     Cached
  o certifi                  2024.12.14  pyhd8ed1ab_0        conda-forge     Cached
  o charset-normalizer            3.4.1  pyhd8ed1ab_0        conda-forge     Cached
  o colorama                      0.4.6  pyhd8ed1ab_1        conda-forge     Cached
  o conda-libmamba-solver        25.1.1  pyhd8ed1ab_0        conda-forge     Cached
  o conda-package-handling        2.4.0  pyh7900ff3_2        conda-forge     Cached
  o conda-package-streaming      0.11.0  pyhd8ed1ab_0        conda-forge     Cached
  o distro                        1.9.0  pyhd8ed1ab_1        conda-forge     Cached
  o h2                            4.2.0  pyhd8ed1ab_0        conda-forge     Cached
  o hpack                         4.1.0  pyhd8ed1ab_0        conda-forge     Cached
  o hyperframe                    6.1.0  pyhd8ed1ab_0        conda-forge     Cached
  o idna                           3.10  pyhd8ed1ab_1        conda-forge     Cached
  o jsonpatch                      1.33  pyhd8ed1ab_1        conda-forge     Cached
  o packaging                      24.2  pyhd8ed1ab_2        conda-forge     Cached
  o platformdirs                  4.3.6  pyhd8ed1ab_1        conda-forge     Cached
  o pluggy                        1.5.0  pyhd8ed1ab_1        conda-forge     Cached
  o pycparser                      2.22  pyh29332c3_1        conda-forge     Cached
  o pysocks                       1.7.1  pyha55dd90_7        conda-forge     Cached
  o requests                     2.32.3  pyhd8ed1ab_1        conda-forge     Cached
  o tqdm                         4.67.1  pyhd8ed1ab_1        conda-forge     Cached
  o truststore                   0.10.0  pyhd8ed1ab_0        conda-forge     Cached
  o urllib3                       2.3.0  pyhd8ed1ab_0        conda-forge     Cached
  o wheel                        0.45.1  pyhd8ed1ab_1        conda-forge     Cached

  Upgrade:
─────────────────────────────────────────────────────────────────────────────────────

  - libcurl                      8.11.1  h332b0f4_0          conda-forge     Cached
  + libcurl                      8.12.1  hc9e6f67_0          pkgs/main        480kB
  - python                       3.12.8  h9e4cc4f_1_cpython  conda-forge     Cached
  + python                       3.13.2  hf623796_100_cp313  pkgs/main         38MB
  - python_abi                     3.12  5_cp312             conda-forge     Cached
  + python_abi                     3.13  0_cp313             pkgs/main          6kB
  - tk                           8.6.13  noxft_h4845f30_101  conda-forge     Cached
  + tk                           8.6.14  h39e8969_0          pkgs/main          4MB

  Downgrade:
─────────────────────────────────────────────────────────────────────────────────────

  - brotli-python                 1.1.0  py312h2ec8cdc_2     conda-forge     Cached
  + brotli-python                 1.0.9  py313h6a678d5_9     pkgs/main        364kB
  - frozendict                    2.4.6  py312h66e93f0_0     conda-forge     Cached
  + frozendict                    2.4.2  py313h06a4308_0     pkgs/main         37kB
  - jsonpointer                   3.0.0  py312h7900ff3_1     conda-forge     Cached
  + jsonpointer                     2.1  pyhd3eb1b0_0        pkgs/main          9kB
  - krb5                         1.21.3  h659f571_0          conda-forge     Cached
  + krb5                         1.20.1  h143b758_1          pkgs/main          1MB
  - libmamba                      2.0.6  h49b8a8d_0          conda-forge     Cached
  + libmamba                      2.0.5  h49b8a8d_1          conda-forge        2MB
  - libmambapy                    2.0.6  py312hbaee817_0     conda-forge     Cached
  + libmambapy                    2.0.5  py313h3a237cd_1     conda-forge      624kB
  - libnghttp2                   1.64.0  h161d5f1_0          conda-forge     Cached
  + libnghttp2                   1.57.0  h2d74bed_0          pkgs/main        690kB
  - libsqlite                    3.48.0  hee588c1_1          conda-forge     Cached
  + libsqlite                    3.46.0  hde9e2c9_0          conda-forge      865kB
  - libuuid                      2.38.1  h0b41bf4_0          conda-forge     Cached
  + libuuid                      1.41.5  h5eee18b_0          pkgs/main         28kB
  - libzlib                       1.3.1  hb9d3cd8_2          conda-forge     Cached
  + libzlib                      1.2.13  h4ab18f5_6          conda-forge       62kB
  - lz4-c                        1.10.0  h5888daf_1          conda-forge     Cached
  + lz4-c                         1.9.4  h6a678d5_1          pkgs/main        159kB
  - mamba                         2.0.6  h8871ed4_0          conda-forge     Cached
  + mamba                         2.0.5  h8871ed4_1          conda-forge      474kB
  - ruamel.yaml                 0.18.10  py312h66e93f0_0     conda-forge     Cached
  + ruamel.yaml                  0.18.6  py313h5eee18b_0     pkgs/main        265kB
  - setuptools                   75.8.0  pyhff2d567_0        conda-forge     Cached
  + setuptools                   72.1.0  py313h06a4308_0     pkgs/main          3MB
  - simdjson                     3.12.0  h84d6215_0          conda-forge     Cached
  + simdjson                     3.11.6  h84d6215_0          conda-forge      245kB

  Summary:

  Install: 13 packages
  Change: 11 packages
  Reinstall: 24 packages
  Upgrade: 4 packages
  Downgrade: 15 packages

  Total download: 107MB

─────────────────────────────────────────────────────────────────────────────────────


Confirm changes: [Y/n] n

[jjerphan]

Might be related to #3126.

[pauldmccarthy]

@jjerphan I just came across that issue :)

To be honest, all of my problems would be solved if there were a way to force conda/mamba/micromamba to only use a single nominated .condarc file, and/or to customise the .condarc search path (e.g. discussed at conda/conda#11838).

My use case is in creating a new root/base environment on a system which is outside of my control, and I want all and any pre-existing configuration files on that system to be ignored, and only the configuration file that I create to be used. Even with the approach I'm currently using (annotating my channel list with #!final, #!top and #!bottom), it is still possible for other channels to be injected into the final channel list.

[holzman]

So, the actual issue AFAICT is that the #!final, #!top, #!bottom annotations aren't supported by mamba (yet).

You could use the behavior from #3126 for your use case (since it ignores any other config files if CONDARC or MAMBARC are set), but #3894 will fix that.

That makes me think though: it would be pretty straightforward to add another variable (EXCLUSIVE_MAMBARC?) that would trigger this behavior on purpose...

[pauldmccarthy]

@holzman Thanks - you may be correct. FWIW I get the same behaviour (packages installed from main) if I remove #!final/#!top/#!bottom from my example.

After a bit more googling, I came across this issue - it seems that conda nominally only supports nodefaults in environment.yml files - it appears that my use of #!final etc to lock down the channel list allowed me to use it in other contexts, when using conda.

I suspect that this will continue to be a problem in light of the relatively recent changes to the Anaconda Terms of Service - I've had several users contact me who work at institutions where all downloads from https://anaconda.org/ have been completely blocked by their system administrators specifically because of the ToS changes (these system administrators are evidently unaware that only packages hosted within https://anaconda.org/main/ are subject to the ToS restrictions).

I agree with the sentiment of @maximlt that nodefaults should always be enforced if it appears in the final resolved channel list, regardless of where/how it was added - this is how I naively understood it to work before digging deeper into the matter.

In any case, I've ended up changing our workflow to use the --rc-file option, e.g. micromamba install --rc-file <my-condarc-file> ..., which seems to do the trick - so this issue is no longer affecting my use of micromamba (and thank heavens I don't have to go back to miniconda/miniforge 😁 ).

[holzman]

I definitely see the effect of #!final (with conda, not mamba). Using your Docker example, with the annotation:

[root@d30c6b5d2a0c /]# cat /miniforge3/.condarc ~/.condarc
channels:      #!final
 - conda-forge #!top
 - nodefaults  #!bottom
channels:
 - defaults
[root@d30c6b5d2a0c /]# /miniforge3/bin/conda config --show channels
channels:
  - conda-forge
  - nodefaults

Without:

[root@d30c6b5d2a0c /]# cat /miniforge3/.condarc ~/.condarc
channels:
 - conda-forge
 - nodefaults
channels:
 - defaults
[root@d30c6b5d2a0c /]# /miniforge3/bin/conda config --show channels
channels:
  - defaults
  - conda-forge
  - nodefaults

Anyhow, as you noticed, using --rc-file will use the rc file specified and ignore the rest.

[pauldmccarthy]

@holzman Aah sorry - it's --rc-file, not --condarc - I've edited my post for posterity.