mamba env export --no-build --from-history changed format in v2 for packages with specified versions

[mathbunnyru]

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

export

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

I installed notebook using: mamba install notebook>=7.2.2
Then exported the environment.

v2:

(base) jovyan@27933482a351:~$ mamba env export --no-build --json --from-history
{
  "channels": [
    "conda-forge"
  ],
  "dependencies": [
    "conda",
    "jupyter_core",
    "jupyterhub-singleuser",
    "jupyterlab",
    "mamba",
    "nbclassic",
    "notebook>=7.2.2",
    "python=3.12"
  ],
  "name": "base",
  "prefix": "/opt/conda"
}

v1:

(base) jovyan@c5da4bba1d79:~$ mamba env export --no-build --json --from-history
{
  "channels": [
    "conda-forge"
  ],
  "dependencies": [
    "jupyter_core",
    "mamba[version='<2.0.0']",
    "python=3.12",
    "jupyterhub-singleuser",
    "jupyterlab",
    "nbclassic",
    "notebook[version='>=7.2.2']",
    "ca-certificates",
    "certifi",
    "openssl"
  ],
  "name": "base",
  "prefix": "/opt/conda"

Note the line with notebook.

v2's output is not compatible with conda's output when some lib was installed specifying the version range

mamba info / micromamba info

No response

Logs

No response

environment.yml

No response

~/.condarc

No response

[mathbunnyru]

There might be tricky cases, I had to change this:
mamba v1: conda-forge::blas[build=openblas]
mamba v2: conda-forge::blas=*

[jjerphan]

Actually, while the current behavior is more consistent (since the --from-history is actually reusing the MatchSpec which have been typed), I think we can consider the change of behavior.

[jjerphan]

@mathbunnyru: I think this is an easy fix you can work on if you are interested.

A priori, one needs to modify this line to parse a string corresponding to the MatchSpec:

mamba/libmamba/src/core/history.cpp

Line 161 in b736741

pkg_specs.push_back(std::move(match_spec));

[jjerphan]

Honestly, I think the current behavior is more consistent.

Note that:

MatchSpec::parse("notebook[version='>=7.2.2']").value().str() == "notebook>=7.2.2"

I do not think this needs to be addressed.

What is your perspective, @mathbunnyru?

[mathbunnyru]

I'm 100% ok with the new output and it looks nice in my opinion.

I think this can be addressed in docs, if there is such one, describing differences of mamba and conda.

[jjerphan]

Yes, let's document it.

[jjerphan]

See: #3697