t18_TransformObufNotPort.cpp File Reference

#include <assert.h>
#include <stdlib.h>
#include <iostream>
#include <sstream>
#include <fstream>
#include <string>
#include <vector>
#include <set>
#include <rpa/platforms.h>
#include <rpa/synchro.h>
#include <rpa/testing.h>
#include <rpa/containers/vector.h>
#include <rpa/containers/set.h>
#include <rpa/containers/row_iterator.h>
#include <rpa/containers/streambuf_tmpnam.h>
#include <rpa/testing_streambuf.h>
#include <rpa/algorithms/transform.h>
#include <rpa/istream_iterator_delimiter.h>

Classes
class	DataIn
	After the template specializations, comes the library. More...
class	DataOut
class	T18In2Out
class	TestRow< OutStrmBuf, RowBufferType >
struct	Tst_T18< TmpNamOutStrmBuf >
	For an output stream class, tests 'transform' with many output buffers types. More...
Defines
#define	DATAIN_BINARY_IO
#define	DATAOUT_BINARY_IO
#define	THR_FACTOR   3
Functions
void	CompareT18 (DataOut *aPtr, size_t aSz, const char *aRangeName, const char *aDelim, std::basic_streambuf< char > &aStreamBuf)
int	main (int aArgC, const char **aArgV)
	Tst_T18.
Variables
static const rpa::thr_nb_t	myNbThr = THR_FACTOR * THR_FACTOR
static rpa::testing_thread	myThrTests [myNbThr]
static const size_t	myMaxElts = 200000 * TST_RATIO
static DataOut	mySample [myMaxElts]
	This contains the output results that each algorithm must calculate.
static DataIn	myInputArr [myMaxElts]
static std::set< DataIn >	myInputSet

---

Define Documentation

#define DATAIN_BINARY_IO

Much faster and convenient than using iostreams. Both methods are possible anyway.

#define DATAOUT_BINARY_IO

Much faster than using iostream for writing data into a file. Both methods are possible anyway.

#define THR_FACTOR   3

The maximum number of threads. In the loop on threads, they are multiplied by three at each loop.

---

Function Documentation

void CompareT18	(	DataOut *	aPtr,
		size_t	aSz,
		const char *	aRangeName,
		const char *	aDelim,
		std::basic_streambuf< char > &	aStreamBuf
	)

It is not necessary to sort the input elements because they are put at initialisation time in a vector which is naturally sorted.

Go back to the beginning of the stream buffer.

This is used for reading the data with the original delimiter.

Go back to the beginning of the stream buffer.

This reads the output result of the algorithm, and puts the elements in a multiset. The data type must be deserializable given the delimiter.

int main	(	int	aArgC,
		const char **	aArgV
	)

Tst_T18.

Tries different types of per-thread buffers. Writes into different sorts of outputs, as long as they are streams. NOTE: We do not attempt to test obuf_iterator with a stream as buffer and a non-ostream_iterator_public output. It may make sense, I do not know yet.

At the end, it is compared against the algorithms output. It is not necessary to sort the input elements because they are given a numerical index.

Fills the containers. The algorithm 'transform' will convert to a DataIn. The content will never be changed.

It is possible to use any kind of input container, but here we concentrate on a plain array.

---

Variable Documentation

DataIn myInputArr[myMaxElts] [static]

These are the input container types. They are very big, and therefore not put on the stack.

std::set< DataIn > myInputSet [static]

const size_t myMaxElts = 200000 * TST_RATIO [static]

Maximum number of input data. The ratio is made for adjusting the testing time.

const rpa::thr_nb_t myNbThr = THR_FACTOR * THR_FACTOR [static]

DataOut mySample[myMaxElts] [static]

This contains the output results that each algorithm must calculate.

rpa::testing_thread myThrTests[myNbThr] [static]

The threads must be global because we have no idea about when they are actually created or stopped. They can be stored in any kind of container as long as threads are not copied nor assigned. This contains a pool of threads we want to use.

---