Description
Summary
Zed scans thousands of files even when they match file_scan_exclusions
, making reference/definition look‑ups noticeably slow.
Description
On a fresh Zed install, typical Python projects with a local .venv and other common build artefacts open fine, but the language analysis subsystem (Pyright + Ruff) quickly balloons to many thousands of files—even those inside paths that are explicitly excluded in file_scan_exclusions
.
Because of this over‑eager indexing / analysis:
- The “Analyzing X/Y files” toast often shows several‑thousand files instead of the few hundred that actually belong to the project.
- Go to Definition, Find References, etc. routinely take 4‑10 s instead of sub‑second.
The issue reproduces both with and without a virtual‑env activated in the integrated terminal.
Steps to reproduce:
- Create a new folder
demo-proj
, add a small Python file (e.g.main.py
withprint("hi")
). - Create a local venv:
python -m venv .venv
or with uv and install something moderately large (pip install numpy
). - Open the folder in Zed latest version (tested on macbook pro m4).
- Observe the file‑analysis counter climbing into the thousands and subsequent sluggish Go to Definition.
Expected Behavior:
Only project‑relevant files (and their stubs) are analyzed; navigation actions complete should be really fast..
Actual Behavior:
Zed traverses and analyzes many files that match the exclusion globs, I assume in the (e.g. .venv/**
, __pycache__/
, .pytest_cache/
) directories, causing seconds‑long delays.
settings.json

Zed Version and System Specs
Zed: v0.190.5 (Zed)
OS: macOS 15.5.0
Memory: 18 GiB
Architecture: aarch64