joaoh82 / rust_sqlite / 25455403967

64%

push

github

feat(bench): harness scaffolding + W1 read-by-PK (SQLR-16, sub-phase 9.1) (#102)

First executable code for the SQLRite-vs-SQLite benchmark suite designed
in #101. Stands up a `benchmarks/` workspace member with a small
`Driver` trait, two engine implementations (SQLRite via `Connection`,
SQLite via `rusqlite`-bundled), one workload end-to-end (W1 read-by-PK),
a criterion entry point, and an aggregator that collapses
`target/criterion/**/estimates.json` into a single results envelope
under `benchmarks/results/`. Excluded from CI alongside `sqlrite-desktop`
+ the SDK cdylibs; runs locally with `make bench`.

Plan-doc:
  - Q1-Q8 resolved on 2026-05-06 and converted to a "Decisions
    (was: open questions)" section, mirroring the post-resolution
    shape of `docs/phase-7-plan.md`. Status moved to
    "approved 2026-05-06 -- 9.1 in flight".
  - Cleaned up the muddled "Not added to the default exclude list"
    sentence in the Repository layout section.

Harness:
  - `Driver` trait is small enough to express every planned workload:
    open / execute / execute_with_params / query_one / query_all.
    `&self` so drivers are cheap to clone into criterion closures.
  - SQLRite driver inlines `?` params via SQL formatting (SQLite-side
    quoting / arity checks unit-tested) -- the engine has no parameter
    binding yet; that lands as a post-9.6 follow-up.
  - SQLite driver runs the Q3-tuned profile at open time
    (WAL + synchronous=NORMAL + 64 MB cache + temp_store=MEMORY) and
    `prepare_cached` on every hot-path SELECT so we measure execution
    cost, not per-iter parse cost.
  - Per-run `TempDir` -- no page-cache / WAL bleed across iterations.
  - Correctness gate: every workload runs `correctness_check` against
    sample keys before timing, so a divergent-result "speedup" can't
    silently land.
  - Q8 versioning baked in: workloads carry a `WorkloadId { id, name,
    version }` and the JSON envelope tags every sample with `W{n}.v{v}`.

Output:
  ... (continued)

0 of 119 new or added lines in 3 files covered. (0.0%)

8704 of 13447 relevant lines covered (64.73%)

1.2 hits per line