I've been writing production .NET since beta 2 -- that is November 2001 for those that don't remember. I've used everything. I've been burned by everything.
EF Core hides what it's doing until it doesn't and then you're reading query plans at 2am wondering why a simple update is taking 400ms. Dapper is honest, but it's also essentially abandoned infrastructure at this point zero database-free test coverage until a PR I submitted last month raised it from 0.61% to 86%.
So I built the thing I actually wanted. pengdows.crud started internally in 2006, ran in production for a decade, and got a full .NET Core 8 rewrite as v1.0. 2.0 just dropped.
pengdows.crud is a SQL-first .NET data access library that gets out of your way. Write the SQL you mean, map results to any shape — tables, DTOs, view models, aggregates — without fighting an ORM's assumptions. Built on ADO.NET with zero provider dependencies, it supports 14 databases including PostgreSQL, SQL Server, Oracle, DuckDB, CockroachDB, and more. Includes built-in connection pool governance, per-call database context for safe singleton use.
