andygrove opened a new pull request, #7545: URL: https://github.com/apache/datafusion-comet/pull/7545
## Which issue does this PR close? Part of #7288. ## Rationale for this change Comet Native Executors run whole Spark queries on a cluster with no JVM in the data plane, and a query that cannot run entirely in the Comet pipeline fails. Running part of a query in Spark, and speaking Spark's own protocols from the cluster, are possible directions for later. This records the design research and the architecture diagram so that the experiments can be discussed. This is split out of https://github.com/apache/datafusion-comet/pull/7289 so that it can be reviewed separately. It is documentation only and does not depend on any code. ## What changes are included in this PR? - A new contributor guide page, `docs/source/contributor-guide/native_executors_research.md`. It gives a short overview of Comet Native Executors and then covers: the places where data crosses between Spark and the cluster in the design, approaches for running part of a query in Spark, a shuffle that both engines can read and write, running JVM expressions on the executors, speaking Spark's control plane and block transfer protocols, a proposed first experiment ([#7384](https://github.com/apache/datafusion-comet/issues/7384)), and when splitting a query pays off. It describes research, not implemented behavior. - The architecture diagram, `docs/source/_static/images/comet-native-executors.svg`, shown on that page. - A toctree entry for the page in `docs/source/contributor-guide/index.md`. ## How are these changes tested? Documentation only, so there are no tests. I checked that `prettier --check` passes on the changed Markdown files, that the SVG is well-formed XML and is covered by the existing `docs/source/_static/images/*.svg` entry in `dev/release/rat_exclude_files.txt`, that the page has one H1 and the ASF license header, that the image path resolves the way the other contributor guide pages reference `_static/images`, and that the issues the page links to exist and match the sentences that cite them. The page makes no claim about code that is not on `main`. I did not build the Sphinx docs. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
