I was copying files from a compressed folder to c:\ using Windows XP SP2 explorer,  it  reported to  take about 87 minutes, but it seems stopped at copying a file so I want to cancel the operation. The "Cancel" button was used but the copying process is still hanging for a while( 3 minutes?), I tried to call "Task manager" but "Ctrl+Alt+Del" does not work. So I shutdown the power and next time Windows start up with Blue Screen Error forever. Even Windows Setup program refuse to continue due to MBR error.

I am building a big workspace and faced link error problem. After two hours research, I noticed that size of one file is 0, which is output file of an idl. I deleted all the idl output files and rebuild the idl file, then the workspaces built successfully.

Installed the alpha version firefox 3.0, no new features found (I am lazy to read its release notes), but what I can tell is that its speed is super fast! http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/firefox-3.0a3pre.en-US.win32.installer.exe I am behind a proxy and I can't sign in hotmail/google via some proxy, this problem also fixed by firefox 3.0.

It is so called a new major release. But the refactoring feature is still very limited -- can only "rename", and the speed is very slow. I created a test project of a simple source file (less than 10 lines), while it took about 2 seconds to "rename".

I am thinking of implementing parameterized TCP protocol over UDP. That is, allow program to configure any parameters they want in order to optimize the network usage. Before I put into action, some work went into my eyes:
http://udt.sourceforge.net/
UDT: UDP-based Data Transfer Protocol
http://www.codeproject.com/internet/udt.asp
A UDP-based Reliable Data Transfer Library

I set include path and lib path of boost path in VC++ IDE, but the build script reset INCLUDE and LIB environment variables before launching the build command. This confused us for a while -- the damn perl script.

I want to override some header file(root\somewhere\a.h) in another directory, so I created a header file (root\aaa\proj\a.h) in the project directory, a source file (root\aaa\b.cpp)  of this project referred this header file, it reports link error -- some method in a.h was not found. This error was fixed by adding ".\" to include path of the project.

I saw a lot of place use CString +=, but I thought this is an inefficient way, so I did the following test:
    int length = 1024 * 16;
    {
        ProfileTimer timer("CString test");
        CString szText;
        for(int i=0; i<length; i++)
        {
            szText += "a";
        }
    }
    {
        ProfileTimer timer("std::string test");
        std::string szText;
        for(int i=0; i<length; i++)
        {
            szText += "a";
        }
    }
    {
        ProfileTimer timer("std::stringstream test");
        std::stringstream ss;
        for(int i=0; i<length; i++)
        {
            ss << "a";
        }
        std::string szText = ss.str();
        szText.c_str();
    }
Build with VC++ 6.0
// Debug configuration
// [3316] CString test costs 0.102501 seconds, CPU cycle:204489240
// [3316] std::string test costs 0.0237083 seconds, CPU cycle:47293355
// [3316] std::stringstream test costs 0.00570771 seconds, CPU cycle:11383169
// Release configuration
// [388] CString test costs 0.0586345 seconds, CPU cycle:116974315
// [388] std::string test costs 0.00277884 seconds, CPU cycle:5540249
// [388] std::stringstream test costs 0.00197818 seconds, CPU cycle:3943002

// Embedding python in C++ program.
// VC2005, boost 1.33.1, Python 2.5
// Reference:
// http://wiki.python.org/moin/boost.python/EmbeddingPython

#include "stdafx.h"
#define _WIN32_WINNT 0x0500
#include <Windows.h>
#include <iostream>
#include <sstream>
#include <boost/python.hpp>
#include <boost/python/converter/registry.hpp>
#pragma comment(lib,"boost_python-vc80-mt-1_33_1.lib")
using namespace boost::python;

void Dump(LPCSTR s)
{
OutputDebugStringA(s);
if( IsDebuggerPresent())
{
OutputDebugStringA("\n");
}
std::cout<<s<<std::endl;
}
void Dump(std::stringstream& ss)
{
Dump(ss.str().c_str());

}
class CppBaseClass
{
public:
virtual void setString(LPCSTR d)
{
std::stringstream ss;
ss<< __FUNCTION__<<"('"<<d<<"')";
Dump(ss);
}
void NonVirtualMethod(int d)
{
std::stringstream ss;
ss<< __FUNCTION__<<"('"<<d<<"')";
Dump(ss);

}
};
class CppClass :public CppBaseClass {
public:
void setString(LPCSTR s)
{
std::stringstream ss;
ss<< __FUNCTION__<<"('"<<s<<"')";
Dump(ss);
}
int getNum() {
return m_data;
}
void setNum(int n)
{
m_data = n;
}
CppClass():m_data(0)
{
InterlockedIncrement(&m_instance_count);
std::stringstream ss;
ss<< __FUNCTION__ << "[" <<reinterpret_cast<int>(this) <<"] in thread "<<GetCurrentThreadId();
Dump(ss);
}
CppClass(int n):m_data(n)
{
InterlockedIncrement(&m_instance_count);
std::stringstream ss;
ss<< __FUNCTION__ << "[" <<reinterpret_cast<int>(this) <<"] in thread "<<GetCurrentThreadId();
Dump(ss);

}
~CppClass()
{
InterlockedDecrement(&m_instance_count);
std::stringstream ss;
ss<< __FUNCTION__ << "["<< reinterpret_cast<int>(this) <<"] in thread "<<GetCurrentThreadId();
Dump(ss);
}
static void Check()
{
std::stringstream ss;
ss<< "Instance count:"<<m_instance_count;
Dump(ss);
}
private:
int m_data;
static long m_instance_count;
};
long CppClass::m_instance_count;

int main( int argc, char ** argv ) {
Py_Initialize();
try {
CppClass cpp;
boost::python::converter::registry::query(type_id<int>());
object main_module((
handle<>(borrowed(PyImport_AddModule("__main__")))));

object main_namespace = main_module.attr("__dict__");
main_namespace["CppClass"] = class_<CppClass>("CppClass")
.def(init<int>())
.def("NonVirtualMethod", &CppBaseClass::NonVirtualMethod)
.def("setString", &CppClass::setString)
.def("setStringBase", &CppBaseClass::setString)
.def("getNum",&CppClass::getNum)
.def("setNum", &CppClass::setNum);
main_namespace["cpp"] = ptr(&cpp);
handle<> ignored(( PyRun_String( "print 'Hello, World'\n"
"cpp = CppClass()\n"
"cpp.NonVirtualMethod(111)\n"
"cpp.setString('pass string test')\n"
"cpp.setStringBase('pass string to base class')\n"
"cpp.setNum(1234)\n"
"print cpp.getNum()\n"
"cpp = CppClass(4)\n"
"cpp = None\n",
Py_file_input,
main_namespace.ptr(),
main_namespace.ptr() ) ));
} catch( error_already_set ) {
PyErr_Print();
}
Py_Finalize();// http://www.boost.org/libs/python/todo.html#pyfinalize-safety
CppClass::Check();
}

_________________________________________________________________
Get the Live.com Holiday Page for recipes, gift-giving ideas, and more.
www.live.com/?addtemplate=holiday

#include <stdio.h>
#include <tchar.h>
#include <windows.h>
#include <iostream>


int CreateFileOfSize(LPCTSTR filename, __int64 len)
{
HANDLE hfile = CreateFile(filename,
GENERIC_WRITE,
FILE_SHARE_READ,
0,
CREATE_ALWAYS,
FILE_ATTRIBUTE_NORMAL,// FILE_ATTRIBUTE_SPARSE_FILE flag will not allocate disk space
0);
if(!hfile)
{
return GetLastError();
}
LARGE_INTEGER li;
li.QuadPart = len;
if( 0 == SetFilePointerEx(hfile, li, 0, FILE_BEGIN) )
{
CloseHandle(hfile);
return GetLastError();
}
if(!SetEndOfFile(hfile) )
{
CloseHandle(hfile);
return GetLastError();
}
CloseHandle(hfile);
return ERROR_SUCCESS;
}

int _tmain(int argc, _TCHAR* argv[])
{
// tested on Windows XP NTFS drive
// verified with diskview.exe (http://www.microsoft.com/technet/sysinternals/FileAndDisk/DiskView.mspx)
int ret = CreateFileOfSize(_T("c:\\test_1g.tmp"), 1024*1024*1024);
std::cout<<"Error code:" << ret<<std::endl;
return 0;
}

---

View Athletes' Collections with Live Search. See it!

Open "Visual Studio 2005 Command Prompt",
change to boost directry
set PYTHON_VERSION="2.5"
set PYTHON_ROOT="c:\python25"
set PYTHON_LIB_PATH="c:\python25\libs"
E:\opensource\boost\boost_1_33_1>bjam "-sTOOLS=vc-8_0" install

import random
import struct
for i in range(0, 5):
    ndict={} #empty dict
    while len(ndict) < 6:
        n = random.randint(1, 33)
        ndict[n] = n
    keys = ndict.keys()
    keys.sort()
    #for k, v in ndict.iteritems():
    red_nums = ' '
    for k in keys:
        print ' ', k
    #print 'len of dict is ', len(ndict)
    n = random.randint(1,16)
    print 'black number is ', n

def get_ext(s): # get extension part of file name
dot_pos = s.rfind('.')
if dot_pos<> -1:
print 'dot position is ', dot_pos, ' input is ', s
s = s[0:dot_pos]
print 'output is ', s
get_ext('a.b') # test the function
get_ext('a.b.c')

I have a long list of constants (all are prefixed with EID_) defined in enum and I want to dump them in user friendly name, so I did a regex replace in Notepad++.
Find what: (EID_[_A-Z0-9]*),
Replace with: case \1: MY_TRACE("PushEvent(\1)"); break;

Solved a runtime problem on corba client today. I have a code snippet that works in one project but failed in another project. After painful investigation and try-and-error investigation, the solution seems to be linking addtional two libraries with the failure project.

Here are performance facts of data persistence I collected for evaluation of some project.
Write about 20,000 fixed sized structures (10 fields) to file, CFile cost 0.02 seconds, file size: 900KB MSXSML cost 3.95 seconds, file size: 5MB Xerces cost 1.01 seconds, file size: 5MB XML via ofstream cost 1.95 seconds, file size: 5MB XML via stringstream cost 10.73 seconds, file size 5MB.
IE consumed 407,816KB memory loading the xml file.