Part VIII. Databases providers for developers

Table of Contents

Getting started
Virtual methods for providers
Synchronous / asynchronous mode
Multi threaded environment
Methods - provider's information
get_name() - mandatory
get_version() - mandatory
get_server_version() - mandatory
supports_feature()
Methods - connection management
open_connection() - mandatory
close_connection() - mandatory
get_database()
Methods - DDL queries
supports_operation()
create_operation()
render_operation()
perform_operation()
Methods - transactions management
begin_transaction()
commit_transaction()
rollback_transaction()
add_savepoint()
rollback_savepoint()
delete_savepoint()
Methods - DML queries
create_parser()
statement_to_sql()
statement_prepare()
statement_execute() - mandatory
handle_async()
Methods - data representation
get_data_handler()
get_def_dbms_type()
escape_string()
unescape_string()
Methods - metadata
Important note about SQL identifiers
Reserved SQL keywords
_info()
_btypes()
schemata() and _schemata()
tables_views() and _tables_views()
columns() and _columns()
constraints_tab() and _constraints_tab()
constraints_ref() and _constraints_ref()
key_columns() and _key_columns()
Methods - misc.
cancel()
create_connection()
is_busy()
Virtual methods for recordsets
Virtual methods for Blob operations
SQL parser
Implementation overview
Generic SQL parser
Provider specific SQL parser
Tips to write a custom parser
GdaSqlStatement — SQL statement
Assembling all the parts
libmain.c
plugin_init()
plugin_get_name()
plugin_get_description()
plugin_get_dsn_spec()
plugin_get_auth_spec()
plugin_create_provider()
Providers' support API
GdaServerProvider — Base class for all the DBMS providers
Subclassing GdaDataSelect — Base class for all the data models returned by DBMS providers when a SELECT statement is executed
GdaPstmt — Base class for prepared statement's
Quark list — Manages lists of KEY=VALUE pairs
SQL rendering API — Adapting the SQL to the database's own SQL dialect
Misc API for database providers — Methods dedicated to implementing providers

For each database engine, Libgda requires an object which maps Libgda's API to the native API offered by that database engine. That object, a database provider needs to inherit GdaServerProvider and implement its virtual methods.

Database provider objects are generally instantiated once by the Libgda framework and can be used several times to open and work on connections to several databases of the same type.

Since Libgda itself is developed in the C language, and that most providers are also implemented in that language, the Libgda library itself contains a set of helper classes and functions to guide you in the addition of a new provider to the GDA framework. Two reference implementations are provided in the providers/skel-implementation directory of Libgda's sources:

  • in providers/skel-implementation/capi: a skeleton implementation from scratch for a database accessed through a C API, where all the methods to implement are partially coded

  • in providers/skel-implementation/models: a skeleton implementation to write a provider for sources which are not databases (which do not implement any SQL) such as the Berkeley DB provider (a BDB database is a key/value collection) or the MDB (MS Access files) provider. This kind of provider "export" tables which are in fact data models.

The helper objects and functions to write database providers are documented in the Providers' support API section.