pqxx::icursorstream Class Reference

Simple read-only cursor represented as a stream of results. More...

#include <cursor.hxx>

List of all members.

Public Types

typedef cursor_base::size_type size_type
typedef
cursor_base::difference_type 
difference_type

Public Member Functions

 icursorstream (transaction_base &context, const std::string &query, const std::string &basename, difference_type sstride=1)
 Set up a read-only, forward-only cursor.
 icursorstream (transaction_base &context, const result::field &cname, difference_type sstride=1, cursor_base::ownershippolicy op=cursor_base::owned)
 Adopt existing SQL cursor. Use with care.
 operator bool () const throw ()
icursorstreamget (result &res)
 Read new value into given result object; same as operator >>.
icursorstreamoperator>> (result &res)
 Read new value into given result object; same as get(result &).
icursorstreamignore (std::streamsize n=1)
 Move given number of rows forward (ignoring stride) without reading data.
void set_stride (difference_type stride)
 Change stride, i.e. the number of rows to fetch per read operation.
difference_type stride () const throw ()

Friends

class icursor_iterator


Detailed Description

Simple read-only cursor represented as a stream of results.

SQL cursors can be tricky, especially in C++ since the two languages seem to have been designed on different planets. An SQL cursor has two singular positions akin to end() on either side of the underlying result set.

These cultural differences are hidden from view somewhat by libpqxx, which tries to make SQL cursors behave more like familiar C++ entities such as iterators, sequences, streams, and containers.

Data is fetched from the cursor as a sequence of result objects. Each of these will contain the number of rows defined as the stream's stride, except of course the last block of data which may contain fewer rows.

This class can create or adopt cursors that live outside any backend transaction, which your backend version may not support.


Member Typedef Documentation


Constructor & Destructor Documentation

pqxx::icursorstream::icursorstream ( transaction_base context,
const std::string &  query,
const std::string &  basename,
difference_type  sstride = 1 
)

Set up a read-only, forward-only cursor.

Roughly equivalent to a C++ Standard Library istream, this cursor type supports only two operations: reading a block of rows while moving forward, and moving forward without reading any data.

Parameters:
context Transaction context that this cursor will be active in
query SQL query whose results this cursor shall iterate
basename Suggested name for the SQL cursor; a unique code will be appended by the library to ensure its uniqueness
sstride Number of rows to fetch per read operation; must be a positive number

References set_stride().

pqxx::icursorstream::icursorstream ( transaction_base context,
const result::field cname,
difference_type  sstride = 1,
cursor_base::ownershippolicy  op = cursor_base::owned 
)

Adopt existing SQL cursor. Use with care.

Forms a cursor stream around an existing SQL cursor, as returned by e.g. a server-side function. The SQL cursor will be cleaned up by the stream's destructor as if it had been created by the stream; cleaning it up by hand or adopting the same cursor twice is an error.

Passing the name of the cursor as a string is not allowed, both to avoid confusion with the other constructor and to discourage unnecessary use of adopted cursors.

Warning:
It is technically possible to adopt a "WITH HOLD" cursor, i.e. a cursor that stays alive outside its creating transaction. However, any cursor stream (including the underlying SQL cursor, naturally) must be destroyed before its transaction context object is destroyed. Therefore the only way to use SQL's WITH HOLD feature is to adopt the cursor, but defer doing so until after entering the transaction context that will eventually destroy it.
Parameters:
context Transaction context that this cursor will be active in
cname Result field containing the name of the SQL cursor to adopt
sstride Number of rows to fetch per read operation; must be a positive number

References set_stride().


Member Function Documentation

icursorstream& pqxx::icursorstream::get ( result res  ) 

Read new value into given result object; same as operator >>.

The result set may continue any number of rows from zero to the chosen stride, inclusive. An empty result will only be returned if there are no more rows to retrieve.

Returns:
Reference to this very stream, to facilitate "chained" invocations ("C.get(r1).get(r2);")

icursorstream & pqxx::icursorstream::ignore ( std::streamsize  n = 1  ) 

Move given number of rows forward (ignoring stride) without reading data.

Returns:
Reference to this very stream, to facilitate "chained" invocations ("C.ignore(2).get(r).ignore(4);")

References pqxx::internal::sql_cursor::move().

pqxx::icursorstream::operator bool (  )  const throw ()

icursorstream& pqxx::icursorstream::operator>> ( result res  ) 

Read new value into given result object; same as get(result &).

The result set may continue any number of rows from zero to the chosen stride, inclusive. An empty result will only be returned if there are no more rows to retrieve.

Returns:
Reference to this very stream, to facilitate "chained" invocations ("C >> r1 >> r2;")

void pqxx::icursorstream::set_stride ( difference_type  stride  ) 

Change stride, i.e. the number of rows to fetch per read operation.

Parameters:
stride Must be a positive number

References pqxx::to_string().

Referenced by icursorstream().

difference_type pqxx::icursorstream::stride (  )  const throw ()


Friends And Related Function Documentation

friend class icursor_iterator [friend]


The documentation for this class was generated from the following files:

Generated on Sat Aug 15 00:16:52 2009 for libpqxx by  doxygen 1.5.8