Add basic assumptions to AGENTS.md#3552
Conversation
| ## Testing | ||
|
|
||
| - Bias towards adding tests to existing files, rather than creating new files. | ||
| - Use existing test fixtures when possible. |
There was a problem hiding this comment.
Should we add a line on our preference for integration testing over mocking?
There was a problem hiding this comment.
Yeah, I mentioned that we should only use mocks if similar tests exist in the codebase. The best place I can think of is in the REST Catalog, but that feels too specific to hardcode in here.
|
Thanks @rambleraptor for extending this, I like it a lot! |
Co-authored-by: Fokko Driesprong <fokko@apache.org>
jayceslesar
left a comment
jayceslesar
left a comment
There was a problem hiding this comment.
It might be good to add something about comparing to upstream implementations when possible (java)
|
|
|
Just pushed up the changes! Thanks for your patience. I've had some OOO time lately. |
4 participants
Rationale for this change
Our AGENTS.md just talks about the security model. We get a non-trivial amount of AI activity on this repo and our AGENTS.md should reflect some basic assumptions we want for those PRs.
A lot of this was inspired by Iceberg AGENTS.md.
This is just meant to be an initial version with some thoughts that I've seen and some of the basic architecture (as seen in the Iceberg AGENTS.md)
Are these changes tested?
Just Markdown.
Are there any user-facing changes?
No.