SQL Databases are the standard tool for manipulating relational data with ACID transactions, but they are not convenient to use from C++. Raw SQL APIs such as SQLite do not match the C++ programming paradigm well: queries are strings parsed at run time, and the API is not type-safe. An ORM or Query Object Model can hide SQL from the C++ code, and improve type safety, but it can result in sub-optimal performance, and adds to complexity.
In many applications, the immense complexity of a SQL database is not at all necessary. Joedb is a minimalist low-level alternative that offers the possibility to:
- incrementally add data to a file in a crash-safe way,
- synchronize data access between multiple processes, on the same machine or over a network connection,
- offer a direct low-level C++ API, with compile-time checking of field names and types.
As shown in the diagram below, joedb compiles the database schema into C++ code. Applications using this code can then manipulate data like a C++ container.
Other systems such as protocol buffers provide similar type-safe mechanisms for serializing structured data, but are missing key properties of a relational database, such as concurrency, transactions, and incremental crash-safe updates. Joedb can offer all these features by storing data as a journal of transactions. The whole data history is stored, so it is possible to re-create any past state of the database. Joedb also has a network protocol, and can operate in a distributed fashion, a bit like git for structured data.
To give an order of magnitude of database complexities, the table below shows
the number of lines of source code of joedb, and some SQL databases. These are
without tests, and measured with grep -c $.
| Database | Lines of code |
|---|---|
| joedb | 18,403 |
| SQLite | 240,533 |
| PostgreSQL | 1,512,558 |
So joedb is an extremely simple low-level foundation for sharing relational data. It would be possible to build a SQL database on top of it, but it is already very convenient as-is.
For more information, please take at a look at the documentation.