Hi, @waahm7. Very appreciated for your interests.

Firstly, you may need to be familiar with the Rust. There are countless materials on the web. If you are nice at C like system language, I think the learning curve is straight. A little more recommendations for old-men-from-c-or-cpp-world: Rust For Systems Programmers, Rust Language Cheat Sheet for quick checking-at-hand.
To get familiar the tools which are used in the Base. If you reach here, we may schedule some remote meeting for some quick project review/help. Feel free to get me at Base slack channel for more realtime feedback.
Understand the basic work flow for Base runtime engine. Just simple:
Input -> SQL Parsing -> IR building ( multi-level abstractions and optimizations, but now it is easy-understanding) -> Kernel codegen (gen executable C code) -> Kernel running (Via a JIT library, now is a blackbox dll) -> Result responding and showing.
For this issue, it is a little easier in that there is a sum aggregation as an example. So, you should firstly try to see if you can understand how the sum aggregation done in Base. The core of sum agg is just an embedded c template.
Here, Base maps the query pipeline semantics into a c semantics directly, which guarantees a top-performance and human-debuggable sql execution. The neat gen output can be directly checked and executed with a standalone c compiler via a simple c runner (current ker_runner.c is just showing the sum agg's gen).
So, to resolve this issue, it is just enough to tweak that template around to gen some suitable c codes according to the operators.

support max/min/avg aggregation #3

Description

Metadata

Metadata

Assignees

Labels

Type

Fields

Projects

Milestone

Relationships

Development

Issue actions