Commit 63f4617b authored by Michal 'vorner' Vaner's avatar Michal 'vorner' Vaner
Browse files

[trac1061] Doxygen comments for database classes

parent 579fd2bf
......@@ -20,17 +20,151 @@
namespace isc {
namespace datasrc {
* \brief Abstract connection to database with DNS data
* This class is defines interface to databases. Each supported database
* will provide methods for accessing the data stored there in a generic
* manner. The methods are meant to be low-level, without much or any knowledge
* about DNS and should be possible to translate directly to queries.
* On the other hand, how the communication with database is done and in what
* schema (in case of relational/SQL database) is up to the concrete classes.
* This class is non-copyable, as copying connections to database makes little
* sense and will not be needed.
* \todo Is it true this does not need to be copied? For example the zone
* iterator might need it's own copy. But a virtual clone() method might
* be better for that than copy constructor.
* \note The same application may create multiple connections to the same
* database. If the database allows having multiple open queries at one
* connection, the connection class may share it.
class DatabaseConnection : boost::noncopyable {
~ DatabaseConnection() { }
virtual std::pair<bool, int> getZone() const;
* \brief Destructor
* It is empty, but needs a virtual one, since we will use the derived
* classes in polymorphic way.
virtual ~ DatabaseConnection() { }
* \brief Retrieve a zone identifier
* This method looks up a zone for the given name in the database. It
* should match only exact zone name (eg. name is equal to the zone's
* apex), as the DatabaseClient will loop trough the labels itself and
* find the most suitable zone.
* It is not specified if and what implementation of this method may throw,
* so code should expect anything.
* \param name The name of the zone's apex to be looked up.
* \return The first part of the result indicates if a matching zone
* was found. In case it was, the second part is internal zone ID.
* This one will be passed to methods finding data in the zone.
* It is not required to keep them, in which case whatever might
* be returned - the ID is only passed back to the connection as
* an opaque handle.
virtual std::pair<bool, int> getZone(const isc::dns::Name& name) const;
* \brief Concrete data source client oriented at database backends.
* This class (together with corresponding versions of ZoneFinder,
* ZoneIterator, etc.) translates high-level data source queries to
* low-level calls on DatabaseConnection. It calls multiple queries
* if necessary and validates data from the database, allowing the
* DatabaseConnection to be just simple translation to SQL/other
* queries to database.
* While it is possible to subclass it for specific database in case
* of special needs, it is not expected to be needed. This should just
* work as it is with whatever DatabaseConnection.
class DatabaseClient : public DataSourceClient {
* \brief Constructor
* It initializes the client with a connection.
* It throws isc::InvalidParameter if connection is NULL. It might throw
* standard allocation exception as well, but doesn't throw anything else.
* \note Some objects returned from methods of this class (like ZoneFinder)
* hold references to the connection. As the lifetime of the connection
* is bound to this object, the returned objects must not be used after
* descruction of the DatabaseClient.
* \todo Should we use shared_ptr instead? On one side, we would get rid of
* the restriction and maybe could easy up some shutdown scenarios with
* multi-threaded applications, on the other hand it is more expensive
* and looks generally unneeded.
* \param connection The connection to use to get data. As the parameter
* suggests, the client takes ownership of the connection and will
* delete it when itself deleted.
DatabaseClient(const std::auto_ptr<DatabaseConnection>& connection);
* \brief Corresponding ZoneFinder implementation
* The zone finder implementation for database data sources. Similarly
* to the DatabaseClient, it translates the queries to methods of the
* connection.
* Application should not come directly in contact with this class
* (it should handle it trough generic ZoneFinder pointer), therefore
* it could be completely hidden in the .cc file. But it is provided
* to allow testing and for rare cases when a database needs slightly
* different handling, so it can be subclassed.
* Methods directly corresponds to the ones in ZoneFinder.
class Finder : public ZoneFinder {
* \brief Constructor
* \param connection The connection (shared with DatabaseClient) to
* be used for queries (the one asked for ID before).
* \param zone_id The zone ID which was returned from
* DatabaseConnection::getZone and which will be passed to further
* calls to the connection.
Finder(DatabaseConnection& connection, int zone_id);
virtual const isc::dns::Name& getOrigin() const;
virtual const isc::dns::RRClass& getClass() const;
virtual FindResult find(const isc::dns::Name& name,
const isc::dns::RRType& type,
isc::dns::RRsetList* target = NULL,
const FindOptions options = FIND_DEFAULT)
const = 0;
DatabaseConnection& connection_;
const int zone_id_;
* \brief Find a zone in the database
* This queries connection's getZone to find the best matching zone.
* It will propagate whatever exceptions are thrown from that method
* (which is not restricted in any way).
* \param name Name of the zone or data contained there.
* \return Result containing the code and instance of Finder, if anything
* is found. Applications should not rely on the specific class being
* returned, though.
virtual FindResult findZone(const isc::dns::Name& name) const;
/// \brief Our connection.
const std::auto_ptr<DatabaseConnection> connection_;
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment