Difference between revisions of "OpenDBX/C API/odbx result"

Latest revision as of 23:14, 28 April 2009

Retrieves a result set from the database server

#include <opendbx/api.h>

int odbx_result(
    odbx_t* handle,
    odbx_result_t** result,
    struct timeval* timeout,
    unsigned long chunk )

Description

Retrieves the result of a query statement sent by odbx_query() from the database server. After a statement was successfully executed, a dynamically allocated object representing the result set is stored in result which must be freed afterwards by odbx_result_finish(). Result sets for SELECT-like statements returned successfully by this function can be processed by calling odbx_row_fetch() until all rows are retrieved. If the statement was an INSERT, UPDATE, DELETE or a similar statement, the number of affected rows is available via odbx_rows_affected().

If a timeout or error occurs, the result pointer is set to NULL. In case of a timeout, odbx_result() should be called again because the query isn't canceled. This function must be called multiple times until it returns zero, even if the query contains only one statement. Otherwise, memory will be leaked and odbx_query() will return an error.

The third parameter (timeout) restricts the time the function is waiting for a result form the server. It may be NULL to wait until a result arrives. Otherwise, it can contain any number of seconds and microseconds in a timeval structure to wait for. The timeval structure must be set each time before calling odbx_result() because its content may get changed by the function. If the server doesn't respond within the timeout, the query isn't canceled! Instead, the next call to this function will wait for the same result set. Waiting the specified time may be implemented in the backends if it is possible, but there is no guarantee. If not, odbx_result() will return not before a responds arrives.

Dependent on the native database library, it may be possible to retrieve all rows at once (if chunk is zero), one by one or more than one row at once. All positive values including zero are allowed as values for chunk If paging (more than one row at once) is not supported by the backend, it will use "one by one" or "all at once" if this is the only option provided.

Return value

odbx_result() returns ODBX_RES_ROWS ("3") if a result set is available and zero if no more results will be returned for the last query string sent via odbx_query() successfully. ODBX_RES_NOROWS ("2") is returned if the statement was executed successfully but will not return a results set (like for INSERT, UPDATE and DELETE statements) and ODBX_RES_TIMEOUT ("1") indicates a timeout. As soon as ODBX_RES_DONE ("0") is returned, the available result sets were processed. The named constants are available since OpenDBX 1.3.2 and the numbers in brackets have to be used instead if a previous release is is the basis for the application development.

It returns an error code whose value is less than zero if one of the operations couldn't be completed successfully. Possible error codes are listed in the error section and they can be feed to odbx_error() and odbx_error_type() to get further details. In this case, the backends are responsible for freeing the data structures they allocated inside the function and to clear the errornous result set if there's one available. The backends will also set the given result pointer to NULL before returning the error. Nevertheless, if a non-fatal error occurred it's possible to retrieve further result sets and this is necessary before calling odbx_query() on the same connection again.

Errors

-ODBX_ERR_BACKEND 

The native database library returned an error

-ODBX_ERR_PARAM 

Either handle or result are NULL or handle is invalid

-ODBX_ERR_NOMEM 

Allocating the required memory for the result failed

-ODBX_ERR_RESULT 

Waiting for a response from the server failed because the connection was lost

See also

• odbx_column_count()
• odbx_column_name()
• odbx_column_type()
• odbx_rows_affected()
• odbx_row_fetch()
• odbx_result_finish()