Community for the Daft project and all things distributed data

Distributed Data Community

<https://github.com/Eventual-Inc/Daft/issues/2969>

Thanks! What kind of use-case are you using Lance for?

Embedding colocation with payload data.

We are implementing Lance persistence first before iceberg given its search and indexing capabilities.

I also noticed that write_lance isn't present in the docs. Should I make another issue for that?

Even better you could make a contribution :wink: 

Check out our /docs folder it should be pretty easy to add it to the appropriate rst!

Would you recommend I fork daft or just create a new branch

It’s pretty ez!

1. Fork the repo
2. Make a change, push it to a branch on your fork
3. Make a PR against the main repo from your fork (you can use the Github UI)

It’s just an additional option in `New Pull Request` — you know how they usually have an option like target branch etc? There will be another option for target repository… Just select the main Eventual-Inc one

Welcome! Looking forward to the contrib :slightly_smiling_face:

<@U07E4G2PMUP>, I think there was a bit of a misunderstanding of how the `write_lance` query gets executed. I replied on the GH issue.

TLDR: we're already using `.to_arrow()` for the write lance operation. The current approach was recommended &amp; reviewed by one of the lancedb core contributors <https://github.com/Eventual-Inc/Daft/pull/2421>

Thanks <@U071FMQUR7W> ! I definitely was caught up on write_df's contents.  Looks like the return type from `lance.fragments.write_fragments()`  is the FragmentMetadata class, which is a catch all for all fragment submodule methods