go-oblast/benchmark, branch v0.10.0

replace Wrap with NewDB/NewConn/NewTx

2026-05-18T21:09:08+00:00

return a concrete type from Wrap() to enable non-Oblast DB operations

2026-05-12T23:13:17+00:00

add PostgreSQL benchmark, comparing lib/pq against pgx both with and w/o Oblast

2026-05-12T21:32:28+00:00

change Handle to a generic interface without explicit dep on database/sql

2026-05-12T11:11:41+00:00

allow both None or ErrNoRows in SelectOne{,Where}

2026-05-08T14:52:46+00:00

The None-returning style is more convenient when the application needs
to genuinely behave differently in this case, e.g. returning 404 from an
HTTP API endpoint (instead of 500 for a generic database error).

The ErrNoRows-returning style is more convenient when control flow is
not different for this case vs. other error cases.

return None instead of ErrNoRows from SelectOne{,Where}

2026-05-04T14:23:24+00:00

fix benchmark calling the wrong test function

2026-04-29T23:19:20+00:00

I should have been suspicious of the Gorm/Gorp numbers matching exactly :)

benchmark: provide a handwritten N=1 implementation for InsertAndDelete

2026-04-29T23:09:48+00:00

As I had expected, this removes an unfair disadvantage for Oblast, which
had to maintain two differently-typed slices for Insert() and Delete().
This is not really a fair comparison since usually you would not be
doing both operations in one go.

For Gorp, this rewrite is neutral. But in a shocking twist, Gorm somehow
does _significantly_ worse with a specialized N=1 implementation:

```
BenchmarkInsertAndDelete/N=1/via_Gorm-24
before: 11508 B/op    157 allocs/op
after:  12840 B/op    172 allocs/op
```

add ctx arguments to most methods

2026-04-29T22:56:38+00:00

benchmark: compare InsertAndDelete to plain SQLite using QueryRow for INSERT

2026-04-29T21:36:28+00:00