Add a new "Air Traffic Data" example

[abey79]

What

This example demonstrates multiple aspects of the Rerun viewer:

• Use of the map view.
• Use of pyproj to transform geospatial data from one coordinate system to another.
• Use GeoPandas to load geospatial data into a 3D view.
• Use Polars to batch data to be sent via rr.send_columns() (use --batch).

pixi run py-build-examples && pixi run -e examples air_traffic_data

Checklist

• I have read and agree to Contributor Guide and the Code of Conduct
• I've included a screenshot or gif (if applicable)
• I have tested the web demo (if applicable):
  • Using newly built examples: app.rerun.io
  • Using examples from latest main build: app.rerun.io
  • Using full set of examples from nightly build: app.rerun.io
• The PR title and labels are set such as to maximize their usefulness for the next release's CHANGELOG
• If applicable, add a new check to the release checklist!

• PR Build Summary
• Docs preview
• Examples preview
• Recent benchmark results
• Wasm size tracking

[github-actions]

Deployed docs

Commit	Link
02b2ab1	https://landing-1pfgfnhvx-rerun.vercel.app/docs

[Wumpf]

@abey79 the reason we don't have this in yet is that it simply is too bad of a user experience, right?

[abey79]

@abey79 the reason we don't have this in yet is that it simply is too bad of a user experience, right?

Yes. It's here to remind us that we need to improve our "lots of entities" story :)

[github-actions]

Latest documentation preview deployed successfully.

Result	Commit	Link
✅	296054b	https://landing-1x8st44fh-rerun.vercel.app/docs

^{Note: This comment is updated whenever you push a commit.}

[github-actions]

Web viewer built successfully. If applicable, you should also test it:

• I have tested the web viewer

Result	Commit	Link
✅	296054b	https://rerun.io/viewer/pr/5449

^{Note: This comment is updated whenever you push a commit.}

[emilk]

I would love to get this merged as a benchmark, even though we're not hitting the benchmark yet.

[abey79]

I would love to get this merged as a benchmark, even though we're not hitting the benchmark yet.

May I interest you in #8163? This is a blocker for the --batch logging strategy of this example. Which is interesting to have along side the "regular" (aka rr.log()) logging pathway for ingestion benchmark (aka why is shuffling data in python faster than our compactor).

Then I can further dust it off and try to push it to mergeable state.

[abey79]

As of today on my machine, using the 2h dataset, I get the following ingestion times:

• with --batch, I get 45s of ingestion time, plus 3-4s of additional latency in the viewer
• without --batch, I get 1min20s of ingestion tine, plus ~35s of additional latency in the viewer

I'm probably missing something, but our batches (SDK-side and viewer-side) should probably do a better job at beating my home-grown, Polars-based batching code.