Skip to content

Split the pipeline by dialect over the shared loop #1199

Description

@peteski22

Problem

_pipeline.py is 4,672 lines, imported by about 30 modules pulling about 80 names. Its interface is wide, so it is a collection of loosely related helpers rather than a module with a small interface.

Proposal

One shared loop and three dialect adapters (chat, messages, responses); the rest of _pipeline.py moves to the domain it serves. Public surface shrinks to a handful of entry points.

Acceptance

  • Importing modules count and imported names recorded before and after.
  • Target under ten names.
  • The integration suite passes.

Depends on

#1198, #1194.

Part of #1171.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

area/refactorCodebase structure improvementstype/tech-debtMaintenance and cleanup

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions