Mamba¶
- ID
mamba- Home page
- Version requirement
>= 2
- Platforms
🐧 Linux · 🍎 macOS · 🪟 Windows
- Operations
installed·outdated·search·install·upgrade·upgrade_all·remove·cleanup- purl types
pkg:mamba- CLI name
mamba- Issues and PRs
- Source
Conda-compatible package manager, reimplemented in C++.
Reaches the same channels conda does, resolving with libsolv
behind a command line of its own rather than delegating to conda. It is
wrapped on the same grounds as nala over apt: same archives,
separate implementation. Since 2.0 it shares no code with conda at all,
the Python executable of the 1.x line having been replaced by a
dynamically linked build of micromamba.
Important
Every operation targets mamba’s currently active environment, which is
base when none is activated. mpm neither activates nor switches
environments: it inspects and mutates whatever mamba resolves from the
inherited CONDA_PREFIX / CONDA_DEFAULT_ENV, exactly as a bare mamba
call in the same shell would, and exactly as the conda wrapper does.
Per-environment targeting is not supported yet.
Caution
Sharing that prefix with conda is why the two are serialized against each other. mamba takes a real lock on the environment and on every package cache directory for the length of a transaction, and conda honors none of them: its own locking covers the repodata cache alone. Running them at once corrupts rather than blocks, which upstream has closed as not planned (conda/conda#13037).
Note
No sync. Nothing in the command set refreshes the index on its own, the
closest being clean --index-cache, which only forces a refetch on the
next operation. The conda wrapper implements none either, so this is parity
rather than a gap.
Warning
mamba upgrade does not exist: unlike conda, mamba never aliased it, and
calling it exits non-zero on an unexpected argument. Upgrades go through
update.
Documentation: mamba user guide.
What mpm adds to mamba¶
Through mpm, mamba gains --extended search, to match against package descriptions.
Bigger still, mpm reaches across every manager at once: mpm installed and mpm outdated cover mamba alongside every other manager you run in one table, mpm upgrade --all updates them together, and mpm sbom exports the whole machine as one bill of materials.
Every mpm command also gains --dry-run and --plan previews, cross-scheme version comparison and purl identifiers. See manager augmentations for how each one is built.
Your mamba commands, in mpm¶
You already know mamba: each operation maps one-to-one onto mpm, in an interface shared by every manager.
To… |
With |
With |
|---|---|---|
List what’s installed |
|
|
List outdated packages |
|
|
Search for a package |
|
|
Install a package |
|
|
Upgrade one package |
|
|
Upgrade everything |
|
|
Remove a package |
|
|
Clear caches |
|
|
Prefix any command above with --dry-run to simulate the underlying manager calls without touching the system: the safe way to watch what mpm would do before trusting it.
Operations¶
Operation |
Supported |
Notes |
|---|---|---|
|
✓ |
|
|
✓ |
|
|
||
|
✓ |
extended search backfilled by |
|
✓ |
|
|
✓ |
|
|
✓ |
|
|
✓ |
|
|
||
|
✓ |
|
|
Selecting and configuring mamba¶
Deselect mamba for a single run with --no-mamba, or persist the choice in your configuration:
[mpm]
mamba = false
Keep it enabled but tune how mpm drives it with a per-manager override:
[mpm.managers.mamba]
timeout = 900
mpm config-template mamba prints every overridable attribute as a ready-to-paste block.
Recipes¶
A few jobs you would otherwise script around mamba, one mpm command each:
Snapshot and clone a machine:
mpm --mamba dump mamba.toml, thenmpm restore mamba.tomlon the next one.Export a compliance SBOM:
mpm --mamba sbom(CycloneDX by default,--spdxfor SPDX).
Privilege escalation¶
mpm runs this manager as the current user and never prepends sudo by default. Flip the policy for its privileged operations with --sudo or the per-manager sudo override.
See privilege escalation for the full policy.
Concurrency¶
mpm never runs mamba at the same time as conda or micromamba: they act on one environment prefix and one package cache, and conda honors none of the locks mamba takes on them. Each mutating operation waits for the previous one, even with a higher --jobs, while managers outside this group keep running in parallel.
Only mutations are held back. The read-only queries (installed, outdated, search) take no backend lock and stay fully concurrent.
Cooldown¶
State of Mamba’s release-age gating, from the cooldown support table:
Status: 🚧 Proposed (library API merged, no CLI flag)
Mechanism:
exclude_newer_timestamp,libmambapyonlyReference: mamba-org/mamba#4254
A cooldown only pays off where a compromised release can be withdrawn while the clock runs, and can only be emulated where the registry dates its releases. From the retraction table:
Registry: anaconda.org / conda-forge (
pkg:conda)Retraction: Relabel: “we do not allow edits or the deletion of packages on conda-forge” (immutability); a bad artifact is labelled broken and “Users will no longer be able to install them by default” (procedure)
Publish date: ✅
upload_time, with thebrokenlabel carried inlabels(API)
With --cooldown set, mpm skips this manager’s install and upgrade operations rather than run them unguarded (fail-closed); --allow-unsupported-managers opts back in.
Version probe¶
The version is probed by running:
$ mamba --version
2.9.0
and extracted with:
r"^(?P<version>\d+\.\d+\.\d+\S*)$"
Reference traces¶
Raw native outputs captured in the manager source: the reference mpm’s parsers were written against. If you know Mamba well and a transcript below looks wrong, or a newer release changed its output format, report it.
$ mamba list --json
{
"log_history": [],
"packages": [
{
"base_url": "https://conda.anaconda.org/conda-forge",
"build_number": 2,
"build_string": "h23cfdf5_2",
"channel": "conda-forge",
"dist_name": "libiconv-1.18-h23cfdf5_2",
"md5": "4d5a7445f0b25b6a3ddbb56e790f5251",
"name": "libiconv",
"platform": "osx-arm64",
"url": "https://conda.anaconda.org/conda-forge/osx-arm64/libiconv-1.18-h23cfdf5_2.conda",
"version": "1.18"
}
]
}
$ mamba update --all --dry-run --json
{
"actions": {
"LINK": [
{
"build": "h23cfdf5_2",
"fn": "libiconv-1.18-h23cfdf5_2.conda",
"name": "libiconv",
"version": "1.18"
}
],
"PREFIX": "/opt/conda",
"UNLINK": [
{
"build": "h23cfdf5_1",
"fn": "libiconv-1.17-h23cfdf5_1.conda",
"name": "libiconv",
"version": "1.17"
}
]
},
"dry_run": true,
"log_history": [],
"prefix": "/opt/conda",
"success": true
}
$ mamba update --all --dry-run --json
{
"dry_run": true,
"log_history": [],
"message": "All requested packages already installed",
"prefix": "/opt/conda",
"success": true
}
Feed any of these through mpm and the raw output becomes one uniform table, the same shape for every manager: filter it, project columns, or export it (mpm --mamba installed --output json, or csv, toml, yaml), each package carrying a purl and a version comparable across managers.
Changelog¶
7.7.0.dev0(unreleased)Implement
mambaandmicromamba, the C++ reimplementation of conda and its statically linked twin. Both read the inventory fromlist --json, accepting the array both bare and inside the envelope2.9.0introduced, and synthesizeoutdatedfrom the dry-run transaction the way thecondawrapper does. Search matches package names exactly without falling back tompm’s own refiltering.Stop running
conda,mambaandmicromambaconcurrently. They act on one environment prefix and one package cache, andcondahonors none of the locksmambatakes on them.