Skip to content

Agent config to instruct where to place a function #1626

@nathanjmcdougall

Description

@nathanjmcdougall

Rather than introducing this linearization logic, let's just use a setting in the Adder object which lets us force a linear pipeline. Better to keep the logic internal to the pipeweld layer.

Originally posted by @nathanjmcdougall in #1595

This is an architectural issue. Before adding a function to a module we should consider multiple options, at least two. Functions should be placed in the module most close to the level of abstraction to which it belongs, not necessarily closest to where to is used. For example, functions which extend a utility module with some follow-up logic to be used in a higher-level module belong in the utility module, not the higher-level module in which it is used. Consult with import linter contracts.

We should update the agent config with these principles

Metadata

Metadata

Labels

agentConfiguration updates for LLM coding agents

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions