Repository navigation
Advanced Download Features #17
Description
The current client supports basic downloads, but several advanced scenarios are missing.
Currently supported downloads include:
- File
- Version of an artifact
- Artifact (downloads latest)
- Group (downloads all latest artifacts)
This issue proposes enhancements for extending download functionality.
- Download of Account
- Downloads all latest artifacts from all groups that the account/user owns
- Should recursively resolve groups under the user’s namespace
- Note: Read feature 2 below before implementing, since version handling applies here too
- Download of specific or multiple versions
- Downloading an artifact, group or user (or account) should use the latest version of an artifact by default
- Introduce a new CLI option (e.g.
--version), for example:- -
-version latestdownloads only the latest artifacts (default) -- version all: downloads all versions of artifacts--version 2025-10-14downloads specific version if available
- -
- When downloading a user, group, or artifact, default to the latest version if not specified
- Download entire Databus
- Add functionality to mirror the entire Databus
- Should respect the same
--versionCLI option for control over versions
Activity
kurzum commentedon Oct 27, 2025
Moves shasum to #28
AggarwalKashvi commentedon Jan 2, 2026
Hi @Integer-Ctrl, I’d like to work on this if it’s still open.
I suggest starting with a scoped change — adding the --version option for downloads (latest / all / specific version) and aligning it with existing behaviour.
I can split this into small PRs if that works. Please assign it to me if okay. Thanks!
Shubham-Padkonde commentedon Oct 27, 2026
I see account downloads are covered by #100 and version-selection work exists in #75. Is the remaining whole-Databus mirror item still wanted, and is there an independent planning scope I could take without duplicating those PRs?
A possible first step would be metadata-only discovery that produces a bounded download plan: enumerate accounts/groups/artifacts, apply the agreed version selection, deduplicate distributions and make interrupted enumeration resumable. It could be tested against a small mock catalogue with pagination and repeated results before any large file transfer is attempted.
If that is a useful next piece, could you delegate the discovery/plan design to me and confirm how it should build on the existing work? I would wait for coordination on the CLI/API shape before implementation.