Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

ADBC driver for DataFusion #2263

Open
tokoko opened this issue Oct 20, 2024 · 7 comments · May be fixed by #2267
Open

ADBC driver for DataFusion #2263

tokoko opened this issue Oct 20, 2024 · 7 comments · May be fixed by #2267
Labels
Type: enhancement New feature or request

Comments

@tokoko
Copy link
Contributor

tokoko commented Oct 20, 2024

What feature or improvement would you like to see?

I started an adbc driver for datafusion in adbc-datafusion repo based on the rust dummy driver in this repo. It's mostly functional with support for get_objects, sql and substrait execution.

I'm not sure what's the most appropriate home for the driver though. Should it go here in arrow-adbc repo, the main datafusion repository or maybe as an independent repository as a datafusion-contrib?

@tokoko tokoko added the Type: enhancement New feature or request label Oct 20, 2024
@tokoko
Copy link
Contributor Author

tokoko commented Oct 20, 2024

fyi @alamb, curious what you think

@lidavidm
Copy link
Member

I think we would like projects to maintain their drivers (as DuckDB does), but the Rust target is also shifting. It's possible we should maintain it in -contrib or in tree until the Rust APIs have settled a bit before trying to upstream it so there is a reasonably stable target to follow.

@lidavidm
Copy link
Member

@mbrobbel maybe something to explore is versioning the Rust API definitions separately as well so there is a very stable target instead of churning the version number every 2 months for no reason

@mbrobbel
Copy link
Contributor

in tree until the Rust APIs have settled a bit

I agree with this. Keeping it in tree helps to understand how changes to core traits impact existing drivers, and drivers must be updated as part of PRs that change these traits.

maybe something to explore is versioning the Rust API definitions separately as well so there is a very stable target instead of churning the version number every 2 months for no reason

I agree it would be better to track this separately. There are a few things to consider:

  • I guess this complicates the release process a little bit?
  • As long as we're still using 0.x.y versions, applying non-breaking minor bumps is trivial for dependents, especially when they are using something like Dependabot.
  • We are going to have to release whenever arrow-rs releases a new major version (we are using arrow-rs types in our public API). Dependents using both arrow-rs and adbc crates are blocked from updating their arrow related dependencies until we release with the latest arrow-rs major version.

@lidavidm
Copy link
Member

lidavidm commented Oct 21, 2024

Ah, I see (wrt arrow-rs types). In that case maybe there's not much value in separating the releases; we can just do minor version bumps.

@tokoko
Copy link
Contributor Author

tokoko commented Oct 21, 2024

thanks, I'll open a PR here then.

@tokoko tokoko linked a pull request Oct 21, 2024 that will close this issue
@github-actions github-actions bot added this to the ADBC Libraries 15 milestone Oct 21, 2024
@tokoko
Copy link
Contributor Author

tokoko commented Oct 21, 2024

Only just now realized I was linking to a private repo in the description 🤦‍♂️ anyway.. PR is up at #2267

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Type: enhancement New feature or request
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants