andygrove commented on code in PR #6734: URL: https://github.com/apache/datafusion-comet/pull/6734#discussion_r4202578542
########## docs/source/contributor-guide/ai_assisted_contributions.md: ########## @@ -0,0 +1,94 @@ +<!-- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +# AI-Assisted Contributions + +Many Comet contributors use AI coding tools to write code, tests and documentation, and to help +review pull requests. That is welcome, and this page sets out what the project expects when you do. + +In short, you are responsible for everything you submit, whichever tool helped you write it. +Automated reviews can be helpful, but a pull request must always be approved by a human. + +## Writing Code with AI Tools + +Reviewers treat you as the author of your pull request, whoever or whatever wrote the code. Before +you open it: + +- Understand the change end to end. Be ready to explain the design and the code in review, and to + justify the choices it makes. +- Point out anything you are unsure about, in the description or in a comment on your own pull + request, so that reviewers know where to look. For example: "this passes the tests, but I don't + know whether it is safe to call from more than one thread." +- Test it. Comet has to return the same results as Spark, including for nulls, overflow, ANSI mode + and time zones, and code that looks right can still get those wrong. Test those cases, and run the + suites that your change affects. [Continuous Integration](ci.md) explains which changes need which + suites. +- Read the diff the way a reviewer will. Remove comments that narrate how the change was developed, + tests that don't cover anything new, and edits unrelated to the change. +- Keep the pull request focused. A tool can produce a large change in minutes, but reviewing it + takes much longer, and small pull requests get reviewed sooner. + +The same goes for issues and comments that you write with a tool's help. They are posted under your +name, so check them before you post them. + +You are also responsible for making sure that your contribution can be licensed under the Apache +License. The ASF's [Generative Tooling Guidance] explains what that involves when a tool wrote part +of it. + +The repository has instructions for coding agents in [`AGENTS.md`] and skills for common Comet +tasks, such as adding an expression or reviewing a pull request, under [`.ai/skills`]. For more on +why understanding your change matters, see DataFusion's [policy on AI-assisted contributions]. + +## Automated Reviews + +A pull request may get an automated review from a bot or a GitHub app, or from an AI agent that a +contributor runs. These reviews can be a useful first pass, and they often find real problems, such +as a missed null case or a result that differs from Spark. + +Treat their findings like any reviewer's comments. If you are the author, check each one, fix the +ones that are real, and reply to the ones that are not with the reason. Don't change your code just +because a tool suggested it. + +If you are reviewing, read what the tool found and how the author responded, and build on it rather +than raising the same points again. Don't treat a clean automated review as evidence that the Review Comment: We should drop that, I think -- 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]
