Enable Repo‑Native Notebook Development with .fabric/ Metadata Folders
(Eliminate Per‑Notebook Downloads)
Description
The current Fabric VS Code extension forces each notebook to be downloaded into its own isolated folder containing Fabric‑specific metadata. This breaks Git‑based workflows, creates duplicated folder structures, and prevents editing notebooks directly from a repo. It also requires repeated downloads and manual publish steps.
Proposal
Introduce a file‑first, repo‑native notebook development model where developers can open and edit notebooks directly from a Git repository. The Fabric extension would store all Fabric‑specific metadata (item IDs, bindings, kernel info, runtime artifacts) in a separate .fabric/ folder that is ignored by Git.
Example Structure
/repo
/notebooks
analysis.ipynb
transform.py
/.fabric
analysis.ipynb/
transform.py/Key Capabilities
- Open any notebook directly from the repository
- Run notebooks on Fabric Spark without downloading
- Store Fabric metadata in .fabric/ (auto‑ignored by Git)
- Remove the need for per‑notebook download folders
- Eliminate the “Publish to Fabric” step
- Allow CI/CD pipelines (fabric‑cicd) to deploy notebooks cleanly
Benefits
- Clean, deterministic, auditable Git workflows
- No duplicated folder structures
- No repeated downloads
- Alignment with industry standards (Databricks Repos, Azure ML, Jupyter)
- Dramatically improved developer experience for enterprise teams
Why This Matters
Fabric notebooks are currently item‑centric, but developers need a file‑centric workflow to integrate with Git, pull requests, and CI/CD. This proposal preserves Fabric’s metadata model while enabling a modern, repo‑native development experience.
