docs: add Understanding Comet Plans user guide page#4086
docs: add Understanding Comet Plans user guide page#4086andygrove merged 7 commits intoapache:mainfrom
Conversation
|
|
||
| | Node | Description | | ||
| | ------------------------ | ----------------------------------------------------------------------------------------------- | | ||
| | `CometScan` | V1 Parquet file scan with native execution. | |
There was a problem hiding this comment.
What are difference between CometScan and CometNativeScan? Some things are in JVM with CometScan?
There was a problem hiding this comment.
I just pushed an update to clarify. CometScan is a hybrid implementation with some back-and-forth between JVM and native code. From 0.15.0, the default is to use CometNativeScan, where execution is fully native.
|
|
||
| As a result, a plan can mix three kinds of nodes: | ||
|
|
||
| - **`Comet*` nodes** that run natively in Rust (for example `CometProject`, |
There was a problem hiding this comment.
it actually makes me think should we call CometNative for native nodes, or rather CometJava for comet nodes running in JVM?
There was a problem hiding this comment.
The LLMs all know about the current names though, so it could be confusing to rename at this point.
Some of the JVM nodes also involve native code, so it is more Hybrid.
Co-authored-by: Oleks V <comphead@users.noreply.github.com>
3 participants
Which issue does this PR close?
Closes #2413.
Rationale for this change
We need to explain to users how to understand Comet plans, covering areas such as:
CometExchanceandCometColumnarExchange.What changes are included in this PR?
New user guide page.
How are these changes tested?
N/A