Workspace caching 2 bench

.github/workflows/bench.yml

@@ -84,9 +84,27 @@ jobs:
       - name: "Extract artifact archive"
         run: tar -xvf benchmarks-walltime.tar
 
+      - name: "Install system dependencies"
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libsasl2-dev libldap2-dev libkrb5-dev
+
+      - name: "Clone airflow workspace"
+        run: |
+          git init airflow
+          git -C airflow remote add origin https://github.com/apache/airflow.git
+          git -C airflow fetch --depth 1 origin 7fa400745ac7aebc7cc4ec21d3a047e9fb258310
+          git -C airflow checkout FETCH_HEAD
+
       - name: "Create venv"
         run: ./target/debug/uv venv --cache-dir .cache
 
+      - name: "Sync airflow workspace"
+        run: ./target/debug/uv sync --directory airflow --cache-dir .cache
+
+      - name: "Prime airflow run"
+        run: ./target/debug/uv run --directory airflow --cache-dir .cache python -V
+
       - name: "Run benchmarks"
         uses: CodSpeedHQ/action@db35df748deb45fdef0960669f57d627c1956c30 # v4.13.1
         with:
@@ -125,6 +143,19 @@ jobs:
           cargo run --bin uv -- pip compile test/requirements/jupyter.in --universal --exclude-newer 2024-08-08 --cache-dir .cache
           cargo run --bin uv -- pip compile test/requirements/airflow.in --universal --exclude-newer 2024-08-08 --cache-dir .cache
 
+      - name: "Clone airflow workspace"
+        run: |
+          git init airflow
+          git -C airflow remote add origin https://github.com/apache/airflow.git
+          git -C airflow fetch --depth 1 origin 7fa400745ac7aebc7cc4ec21d3a047e9fb258310
+          git -C airflow checkout FETCH_HEAD
+
+      - name: "Sync airflow workspace"
+        run: cargo run --bin uv -- sync --directory airflow --cache-dir .cache
+
+      - name: "Prime airflow run"
+        run: cargo run --bin uv -- run --directory airflow --cache-dir .cache python -V
+
       - name: "Build benchmarks"
         run: cargo codspeed build --profile profiling -p uv-bench
 

crates/uv-bench/Cargo.toml

@@ -23,8 +23,15 @@ name = "uv"
 path = "benches/uv.rs"
 harness = false
 
+[[bench]]
+name = "airflow_run"
+path = "benches/airflow_run.rs"
+harness = false
+
 [dev-dependencies]
+uv = { path = "../uv", default-features = false }
 uv-cache = { workspace = true }
+uv-cli = { workspace = true }
 uv-client = { workspace = true }
 uv-configuration = { workspace = true }
 uv-dispatch = { workspace = true }
@@ -43,9 +50,13 @@ uv-types = { workspace = true }
 uv-workspace = { workspace = true }
 
 anyhow = { workspace = true }
+clap = { workspace = true }
 criterion = { version = "4.0.3", default-features = false, package = "codspeed-criterion-compat", features = ["async_tokio"] }
+fs-err = { workspace = true }
 jiff = { workspace = true }
+reqwest = { workspace = true }
 tokio = { workspace = true }
+toml = { workspace = true }
 
 [package.metadata.cargo-shear]
 ignored = ["uv-extract"]

crates/uv-bench/benches/airflow_run.rs

@@ -0,0 +1,74 @@
+//! Benchmark `uv run python -V` in a warm cache against a fully synced airflow workspace.
+//!
+//! Lives in its own benchmark binary (separate process from `benches/uv.rs`) so that the
+//! resolver benchmarks don't pollute the process state before [`uv::run`] is first invoked.
+
+use std::hint::black_box;
+
+use clap::Parser;
+use criterion::{Criterion, criterion_group, criterion_main, measurement::WallTime};
+use uv_cli::Cli;
+
+#[expect(clippy::print_stderr)]
+fn run_noop_airflow(c: &mut Criterion<WallTime>) {
+    let airflow_dir = std::path::absolute("../../airflow").unwrap();
+    if !airflow_dir.join("uv.lock").exists() {
+        let commit = "7fa400745ac7aebc7cc4ec21d3a047e9fb258310";
+        let repo = "https://github.com/apache/airflow.git";
+        eprintln!(
+            "Airflow checkout does not exist, not running benchmark.\n\
+            To set up:\n\
+            git init ../airflow\n\
+            git -C ../airflow remote add origin {repo}\n\
+            git -C ../airflow fetch --depth 1 origin {commit}\n\
+            git -C ../airflow checkout FETCH_HEAD"
+        );
+        return;
+    }
+    let cache_dir = std::path::absolute("../../.cache").unwrap();
+
+    let runtime = tokio::runtime::Builder::new_current_thread()
+        .enable_all()
+        .build()
+        .unwrap();
+
+    let airflow_dir = airflow_dir.to_string_lossy().to_string();
+    let cache_dir = cache_dir.to_string_lossy().to_string();
+
+    // First call initializes `uv_flags`, logging, miette, etc. Subsequent calls reuse them.
+    let cli = Cli::try_parse_from([
+        "uv",
+        "run",
+        "--directory",
+        &airflow_dir,
+        "--cache-dir",
+        &cache_dir,
+        "--quiet",
+        "python",
+        "-V",
+    ])
+    .unwrap();
+    runtime.block_on(uv::run(cli, true)).unwrap();
+
+    c.bench_function("run_noop_airflow", |b| {
+        b.iter(|| {
+            let cli = Cli::try_parse_from([
+                "uv",
+                "run",
+                "--directory",
+                black_box(&airflow_dir),
+                "--cache-dir",
+                black_box(&cache_dir),
+                "--quiet",
+                "--offline",
+                "python",
+                "-V",
+            ])
+            .unwrap();
+            runtime.block_on(uv::run(cli, false)).unwrap();
+        });
+    });
+}
+
+criterion_group!(airflow_run, run_noop_airflow);
+criterion_main!(airflow_run);

crates/uv-bench/benches/uv.rs

@@ -6,7 +6,7 @@ use uv_cache::Cache;
 use uv_client::{BaseClientBuilder, Connectivity, RegistryClientBuilder};
 use uv_distribution_types::Requirement;
 use uv_python::PythonEnvironment;
-use uv_resolver::Manifest;
+use uv_resolver::{Lock, Manifest};
 
 fn resolve_warm_jupyter(c: &mut Criterion<WallTime>) {
     let manifest = Manifest::simple(vec![Requirement::from(
@@ -47,11 +47,43 @@ fn resolve_warm_airflow(c: &mut Criterion<WallTime>) {
 //     c.bench_function("resolve_warm_airflow_universal", |b| b.iter(&run));
 // }
 
+/// Benchmark lock file parsing for the airflow workspace.
+fn parse_airflow_lockfile(c: &mut Criterion<WallTime>) {
+    let lockfile = airflow_lockfile();
+
+    c.bench_function("parse_airflow_lockfile", |b| {
+        b.iter(|| toml::from_str::<Lock>(black_box(&lockfile)).unwrap());
+    });
+}
+
+/// Fetch the airflow lockfile from GitHub, caching it locally.
+fn airflow_lockfile() -> String {
+    let cache_path = std::path::absolute("../../.cache/airflow.uv.lock").unwrap();
+    if let Ok(contents) = fs_err::read_to_string(&cache_path) {
+        return contents;
+    }
+
+    let url = "https://raw.githubusercontent.com/apache/airflow/7fa400745ac7aebc7cc4ec21d3a047e9fb258310/uv.lock";
+    let error_message = "Failed to fetch airflow lockfile from GitHub";
+    let response = reqwest::blocking::get(url)
+        .expect(error_message)
+        .error_for_status()
+        .expect(error_message)
+        .text()
+        .expect(error_message);
+
+    fs_err::create_dir_all(cache_path.parent().unwrap()).unwrap();
+    fs_err::write(&cache_path, &response).unwrap();
+
+    response
+}
+
 criterion_group!(
     uv,
     resolve_warm_jupyter,
     resolve_warm_jupyter_universal,
-    resolve_warm_airflow
+    resolve_warm_airflow,
+    parse_airflow_lockfile,
 );
 criterion_main!(uv);
 

crates/uv-distribution/src/metadata/dependency_groups.rs

@@ -80,7 +80,6 @@ impl SourcedDependencyGroups {
             } else {
                 MemberDiscovery::None
             },
-            ..DiscoveryOptions::default()
         };
 
         // The subsequent API takes an absolute path to the dir the pyproject is in

crates/uv-distribution/src/metadata/requires_dist.rs

@@ -65,7 +65,6 @@ impl RequiresDist {
             } else {
                 MemberDiscovery::None
             },
-            ..DiscoveryOptions::default()
         };
         let Some(project_workspace) =
             ProjectWorkspace::from_maybe_project_root(install_path, &discovery, cache).await?

crates/uv-workspace/src/lib.rs

@@ -1,7 +1,6 @@
 pub use workspace::{
-    DiscoveryOptions, Editability, MemberDiscovery, ProjectDiscovery, ProjectWorkspace,
-    RequiresPythonSources, VirtualProject, Workspace, WorkspaceCache, WorkspaceError,
-    WorkspaceMember,
+    DiscoveryOptions, Editability, MemberDiscovery, ProjectWorkspace, RequiresPythonSources,
+    VirtualProject, Workspace, WorkspaceCache, WorkspaceError, WorkspaceMember,
 };
 
 pub mod dependency_groups;