Add cache garbage collection

Cargo.toml

@@ -73,6 +73,8 @@ pretty_assertions = "1.4.0"
 proptest = "1.3.1"
 pulldown-cmark = { version = "0.9.3", default-features = false }
 rand = "0.8.5"
+regex = "1.9.3"
+rusqlite = { version = "0.29.0", features = ["bundled"] }
 rustfix = "0.6.1"
 same-file = "1.0.6"
 security-framework = "2.9.2"
@@ -162,6 +164,8 @@ pasetors.workspace = true
 pathdiff.workspace = true
 pulldown-cmark.workspace = true
 rand.workspace = true
+regex.workspace = true
+rusqlite.workspace = true
 rustfix.workspace = true
 semver.workspace = true
 serde = { workspace = true, features = ["derive"] }

benches/README.md

@@ -9,7 +9,23 @@ cd benches/benchsuite
 cargo bench
 ```
 
-The tests involve downloading the index and benchmarking against some
+However, running all benchmarks would take many minutes, so in most cases it
+is recommended to just run the benchmarks relevant to whatever section of code
+you are working on.
+
+## Benchmarks
+
+There are several different kinds of benchmarks in the `benchsuite/benches` directory:
+
+* `global_cache_tracker` — Benchmarks saving data to the global cache tracker
+ database using samples of real-world data.
+* `resolve` — Benchmarks the resolver against simulations of real-world workspaces.
+* `workspace_initialization` — Benchmarks initialization of a workspace
+ against simulations of real-world workspaces.
+
+### Resolve benchmarks
+
+The resolve benchmarks involve downloading the index and benchmarking against some
 real-world and artificial workspaces located in the [`workspaces`](workspaces)
 directory.
 
@@ -21,15 +37,32 @@ faster. You can (and probably should) specify individual benchmarks to run to
 narrow it down to a more reasonable set, for example:
 
 ```sh
-cargo bench -- resolve_ws/rust
+cargo bench -p benchsuite --bench resolve -- resolve_ws/rust
 ```
 
 This will only download what's necessary for the rust-lang/rust workspace
 (which is about 330MB) and run the benchmarks against it (which should take
 about a minute). To get a list of all the benchmarks, run:
 
 ```sh
-cargo bench -- --list
+cargo bench -p benchsuite --bench resolve -- --list
+```
+
+### Global cache tracker
+
+The `global_cache_tracker` benchmark tests saving data to the global cache
+tracker database using samples of real-world data. This benchmark should run
+relatively quickly.
+
+The real-world data is based on a capture of my personal development
+environment which has accumulated a large cache. So it is somewhat arbitrary,
+but hopefully representative of a challenging environment. Capturing of the
+data is done with the `capture-last-use` binary, which you can run if you need
+to rebuild the database. Just try to run on a system with a relatively full
+cache in your cargo home directory.
+
+```sh
+cargo bench -p benchsuite --bench global_cache_tracker
 ```
 
 ## Viewing reports

benches/benchsuite/Cargo.toml

@@ -11,8 +11,10 @@ publish = false
 
 [dependencies]
 cargo.workspace = true
+cargo-util.workspace = true
 criterion.workspace = true
 flate2.workspace = true
+rand.workspace = true
 tar.workspace = true
 url.workspace = true
 
@@ -26,3 +28,7 @@ harness = false
 [[bench]]
 name = "workspace_initialization"
 harness = false
+
+[[bench]]
+name = "global_cache_tracker"
+harness = false