sqlite_orm v1.0 Release Notes

Release Date: 2018-03-09 // over 3 years ago
  • ๐Ÿš€ Finally sqlite_orm v1.0 is released. This is a first stable version. All future versions with the same major version number will have back-compatibility with this version. If you meet broken compatibility within the same major version please report a bug in the issues section.

    ๐Ÿ”‹ Features

    • No raw string queries :
      forget about db << "SELECT " + idColumnName + " FROM " + myTableName + " WHERE " + idColumnName + " < 10 ORDER BY " + nameColumnName;. Just write storage.select(&Object::id, where(c(&Object::id) < 10), order_by(&Object::name)); instead
    • Intuitive syntax
      most of SQLite3 keywords are provided as functions in sqlite_orm: foreign_key, unique, autoincrement, default_value, collate, using_, on, cross_join, natural_join, left_join, join, left_outer_join, inner_join, offset, limit, is_null, is_not_null, in, where, order_by, group_by, between, like, date, datetime, char_, trim, ltrim, rtrim, changes, length, abs, lower, upper, avg, count, sum, max, min, total, group_concat, distinct, all, rowid, oid, _rowid_. Just imagine SQL syntax is provided in your IDE. (Hint: don't forget to add using namespace sqlite_orm)
    • Comfortable interface - one code line per single query
      yes, there are no service objects required to be declared to make a single query. You can write the most complicated queries within a single semicolon:

      storage.select(columns(&Visit::mark, &Visit::visited_at, &Location::place), inner_join<Location>(on(is_equal(&Visit::location, &Location::id))), where(is_equal(&Visit::user, id) and greater_than(&Visit::visited_at, fromDate) and lesser_than(&Visit::visited_at, toDate) and lesser_than(&Location::distance, toDistance)), order_by(&Visit::visited_at));

    • Built with modern C++14 features (no macros and external scripts)
      yes, some ORM libs require scripts/macros to make columns to members mapping work. But sqlite_orm just works as is

    • ๐Ÿ‘ CRUD support
      โšก๏ธ declare a variable and insert it without any other unnecessary stuff. Next get it by a primary key, update it, remove it or replace it.

    • ๐Ÿ‘ Pure select query support
      also you can just get a std::vector (or any other STL-compatible container) of any column with or without any desired where conditions. Or you can even select several column in a vector of tuples.

    • STL compatible
      it means two things: 1) select your objects or columns in any STL compatible container (std::list, QList or even nlohmann::json); 2) iterate your objects in C++11 for loop:

      for(auto &user : storage.iterate<User>()) { cout << storage.dump(user) << endl; }

    • ๐Ÿ‘ Custom types binding support
      sqlite_orm understands implicitly what column type must be by member pointer type you provide. E.g. std::string member pointer maps to TEXT, int, long map to INTEGER, float, double map to REAL. But you can also use your custom types if you have it. You can even bind your enum to be mapped as string or int or whatever. Or even bind boost::optional<T> as nullable generic type.

    • ๐Ÿ‘ BLOB support
      BLOB SQLite type maps to std::vector<char> or you can add binding to any other type.

    • ๐Ÿ‘ FOREIGN KEY support
      you can use FOREIGN KEY with intuitive syntax: foreign_key(&Visit::location).references(&Location::id). And you don't need to call PRAGMA foreign_keys = 1 every time - storage class calls it for you on every database open if there is at least one foreign key exists.

    • ๐Ÿ‘ Composite key support
      ๐Ÿ‘ PRIMARY KEY with several columns also supported. Just write primary_key(&User::id, &User::firstName) and your composite key is ready to go.

    • ๐Ÿ‘ JOIN support
      ๐Ÿ‘ all kinds of JOIN supported by SQLite are also supported by the lib.

    • ๐Ÿ‘ Transactions support
      ๐ŸŽ transaction is one the most important performance improvement tool. There are three different ways to use transactions in the lib to make your code more flexible and stable.

    • Migrations functionality
      ๐Ÿ”€ sometimes when you use some ORM libs you need to create your database with tables first. Someone performs it at runtime, someone creates all tables with a SQLite client and adds this file in the project assets. Forget about it. Just call sync_schema and storage will check all tables and columns and if there is something missing it will recreate/alter it. sync_schema guarantees that schema will be the same as you specified during make_storage call.

    • Powerful conditions
      don't be shy - use any combinations of conditions during selection/deleting.

    • ๐Ÿ‘ INDEX support
      ๐Ÿ‘‰ use indexes as is - just specify member pointer in make_index function.

    • Follows single responsibility principle
      this is a very important thing which many developers omit - your data model classes must know nothing about storage and other services. It is very useful if your software has a lot of modules and sometimes you change some of them.

    • Easy integration
      single header, no .cpp files. Use conan, cmake or just include it as is.

    • The only dependency
      5 seconds required to connect the lib to your project.

    • ๐Ÿ’… C++ standard code style
      no 'initcapped' C# like function names, no camel case in public function/classes names. Include it and use it just like it is one of the standard headers.

    • No undefined behaviour
      as you know some code in standard library can produce undefined behaviour (e.g. std::vector<int>()[5]). sqlite_orm creators do not like undefined behavior at all. So if something goes wrong be ready to catch std::system_error. By this std::system_error you can know whether error happened in SQLIte or in the lib by inspecting the error_category.

    sqlite_orm.h SHA1 6e0b40c2b7122c02cb6d9efbade487689d933827