BlueStrait - My Life, My Site - 程序员

DBPROP_COMMANDTIMEOUT does not seem to work?

bluestrait@gmail.com(黄敏) — Thu,03 Jun 2010 02:29:17 +0800

This is set as a property to a dbcommand,

 程序代码

propSet.AddProperty(DBPROP_COMMANDTIMEOUT, 30);

But the command does not seem to work after 30 seconds.

This post appears making sense to me:

引用内容

o The timeout no longer counts, once the client library have received some
data back from SQL Server. (This is something I have read in a book; I
have not verified it myself.)
o If the timeout elapses when SQL Server has been inserting data into a
table for 30 seconds, or even worse been updating rows row 30 seconds,
it will not yield on the spot. The effect of the currently executing
statement must be rolled back. And rollbacks usually takes longer than
command that started the operation.

In my case, the first one makes sense because the query in my test command does return 696164 rows. That explains why the command took around 175 seconds and did not report any timeout failure. After all, if there is data incoming, why the command can be assumed hanged?

If I put a sleep in a sproc, the timeout will still occur, obviously since no data is returned to the client, the client think the command is stalled.

Further reading:
This MSDN blog explains query timeouts are a client-side concept only.

But this is probably the best explanation about how the timeout is calculated: http://msdn.microsoft.com/en-us/library/system.data.sqlclient.sqlcommand.commandtimeout.aspx

引用内容

This property is the cumulative time-out for all network reads during command execution or processing of the results. A time-out can still occur after the first row is returned, and does not include user processing time, only network read time.

[zt] Virtuality

bluestrait@gmail.com(黄敏) — Mon,24 May 2010 19:03:22 +0800

A very insightful article about Virtuality in C++:

http://www.gotw.ca/publications/mill18.htm

static variable in header file?

bluestrait@gmail.com(黄敏) — Wed,19 May 2010 19:54:00 +0800

A few discussions:

http://stackoverflow.com/questions/92546/variable-declarations-in-header-files-static-or-not

This confirms that a static variable declared in a header file will leave one copy of it created for each source file in which the header file is included.

typedef or class wrapper

bluestrait@gmail.com(黄敏) — Tue,18 May 2010 20:07:55 +0800

I faced a question to ensure correct parameter type is passed in, although the type essentially is a WORD only:
1. typedef

 程序代码

typedef WORD ParamType

Pro: no run-time overhead cost
Con: no compiler checking hence the app can actually pass in a WORD, or int or whatever numeric values that can be cast into WORD.

2. or use a class wrapper:

 程序代码

class CParamType
{
public:
    CParamType(WORD wValue) : m_wValue(wValue) {};
    virtual ~CParamType() : {};

    WORD m_wValue;
}

Pro: compiler checking will kick in to enforce valid CParamType only.
Con: extra class means a performance hit.

Which way I should go?

Some discussion:
1. http://www.cplusplus.com/forum/general/8423/
Basically use the class when in a special compiler mode to be able to detect the error but fall back to typedef in other compiler mode.

2. http://stackoverflow.com/questions/376452/enforce-strong-type-checking-in-c-type-strictness-for-typedefs
It seems BOOST_STRONG_TYPEDEF from boost library is recommended by many. It is a macro introduced in boost, and internally uses struct to differentiate between types. Essentially same as using a simple class wrapper listed above.

va_start does not work with reference parameter

bluestrait@gmail.com(黄敏) — Thu,06 May 2010 21:48:20 +0800

It is the first time I noticed this issue.

Basically the first parameter of a unknown number of parameters function call must have the first parameter passed as value.

See the full discussion at http://stackoverflow.com/questions/222195/are-there-gotchas-using-varargs-with-reference-parameters

Compiler expression evaluation Interesting

bluestrait@gmail.com(黄敏) — Mon,03 May 2010 02:18:43 +0800

 程序代码

    CString strTest2(_T("Hello")), strTest3;
    LPCTSTR szBuf = TRUE ? NULL : strTest2;
    LPCTSTR szBuf2 = TRUE ? NULL : strTest3;
    LPCWSTR szBuf3 = TRUE ? (LPCTSTR)NULL : strTest3;
    LPCWSTR szBuf4 = TRUE ? NULL : (LPCTSTR)strTest3;

What will be the value of these buffer pointers?

ANSWER:
szBuf --> Same as m_pszData of strTest3
szBuf2 --> Same as m_pszData of strTest3
szBuf3 --> NULL
szBuf4 --> NULL

Explanation:
It must have something to do with the type implication here.

Building Boost.Locale

bluestrait@gmail.com(黄敏) — Mon,19 Apr 2010 02:04:07 +0800

1. download and install cmake
2. download and uncompress icu 4.4 binary format
3. download and uncompress boost 1.33.1+.
3. download boost locale library source code and uncompress.
4. create build directory:

 程序代码

mkdir boost_locale_build
cd boost_locale_build

5. set up environment variable for cmake and then run cmake

 程序代码

set CMAKE_INCLUDE_PATH=D:\devel\Lab\icu\include;D:\devel\mhuangTfs\3rdParty\Boost\1.35\main
set CMAKE_LIBRARY_PATH=D:\devel\Lab\icu\lib
cmake ..\boost_locale\libs\locale

This will create .vcproj files under boost_locale_build directory

D:\devel\Lab\boost_locale_build>dir
Volume in drive D has no label.
Volume Serial Number is B6AD-2046

Directory of D:\devel\Lab\boost_locale_build

19/04/2010  03:53 PM              .
19/04/2010  03:53 PM              ..
19/04/2010  03:53 PM            16,229 ALL_BUILD.vcproj
19/04/2010  03:53 PM            17,919 boost_locale-static.vcproj
19/04/2010  03:53 PM            20,828 boost_locale.vcproj
19/04/2010  03:53 PM            19,503 boundary.vcproj
19/04/2010  03:53 PM            19,503 calendar.vcproj
19/04/2010  03:53 PM            15,364 CMakeCache.txt
19/04/2010  03:53 PM              CMakeFiles
19/04/2010  03:53 PM             7,090 cmake_install.cmake
19/04/2010  03:53 PM            19,481 collate.vcproj
19/04/2010  03:53 PM            19,569 conversions.vcproj
19/04/2010  03:53 PM             1,063 CTestTestfile.cmake
19/04/2010  03:53 PM            19,437 hello.vcproj
19/04/2010  03:53 PM            16,587 INSTALL.vcproj
19/04/2010  03:53 PM            25,239 Project.sln
19/04/2010  03:53 PM            16,587 RUN_TESTS.vcproj
19/04/2010  03:53 PM            19,609 test_boundary.vcproj
19/04/2010  03:53 PM            19,609 test_codepage.vcproj
19/04/2010  03:53 PM            19,587 test_collate.vcproj
19/04/2010  03:53 PM            19,587 test_convert.vcproj
19/04/2010  03:53 PM            19,631 test_date_time.vcproj
19/04/2010  03:53 PM            19,653 test_formatting.vcproj
19/04/2010  03:53 PM            19,631 test_generator.vcproj
19/04/2010  03:53 PM            19,609 test_ios_prop.vcproj
19/04/2010  03:53 PM            19,587 test_message.vcproj
19/04/2010  03:53 PM            19,525 wboundary.vcproj
19/04/2010  03:53 PM            19,591 wconversions.vcproj
19/04/2010  03:53 PM            19,459 whello.vcproj
19/04/2010  03:53 PM            15,562 ZERO_CHECK.vcproj
              27 File(s)        485,039 bytes
               3 Dir(s)  33,357,881,344 bytes free

6. After that, The IDE or MSBuild of MSVC 2008 can be used to build the relevant projects.

Basic GNU Gettext how-to

bluestrait@gmail.com(黄敏) — Wed,14 Apr 2010 20:28:38 +0800

1. Mark the strings using chosen keywoard (for example, '_')

2. Extract text template out of source code with specified keyword and domain name:

 程序代码

\gettext-0.17\bin\xgettext -k_ PnxErrors.h Error.cpp -ophoenix.po -dPhoenix

3. Creating a po file out of template with specified locale so that it can be translated.

 程序代码

\gettext-0.17\bin\msginit -iphoenix.po -ophoenix.zh_CN.po -lzh_CN

this step will produce phoenix.zh_CN.po file under current directory.

4. Translate the strings in produced zh_CN.po. Make sure the file is saved with UTF-8 encoding, and the charset is set to UTF-8, as shown in example below:

 程序代码

"Content-Type: text/plain; charset=UTF-8\n"

5. After translation, build the binary string resource file (.bo):

 程序代码

\gettext-0.17\bin\msgfmt -ozh_CN\LC_MESSAGES\phoenix.mo phoenix.zh_CN.po

6. Later if the source code has been updated and new strings added, step 2 can be run again to create updated string template. After that, msgmerge can be run to update the existing translation files to add in the new strings:

 程序代码

\gettext-0.17\bin\msgmerge.exe -U phoenix.zh_CN.po phoenix.en_US.po

Note: it seems it is not possible to remove an existing translation if a string does not exist any more in the source code.

NOTE:
1. SetThreadLocale() controls which language gettext will use.
2. However, environment variables such as LANG, LANGUAGE, LC_ALL takes precendence of SetThreadLocale() if any of them is present present, unfortunately.
3. User locale is taken if neither thread locale nor environment variables set.
This is actually a big disadvantage of gettext tool which may prevents of the use of this tool in a server environment, when the requirements is to provide localized data to each client based on their locale. SetThreadLocale() is definitely a no-go.

l in Format Specification Fields

bluestrait@gmail.com(黄敏) — Tue,13 Apr 2010 21:07:11 +0800

It is the first time I noticed that you can use l as a size specification for a string parameter, for example:

 程序代码

    CStringA strTest;
    strTest.Format("%ls", L"Hello");

It means the parameter is a wide-character string. It works for both printf and wprintf.

I only realized it after I did unnecessary W2A conversion on a wide-char string which led to ASSERT error from ATL and crashed the app. The guilty code is like below:

 程序代码

    CStringA strTest;
    strTest.Format("%ls", "Hello");

Reference: http://msdn.microsoft.com/en-us/library/tcxf1dw6(VS.90).aspx

Evaluating the performance of a stored proc

bluestrait@gmail.com(黄敏) — Wed,10 Mar 2010 17:53:04 +0800

SET STATISTICS TIME { ON | OFF }

SET STATISTICS IO { ON | OFF }
can be used to help do the IO count.

References:
SET STATISTICS TIME (Transact-SQL)
SET (Transact-SQL)

Differences in behavior between SQLNCLI and MDAC

bluestrait@gmail.com(黄敏) — Thu,04 Feb 2010 22:13:21 +0800

Following differences are listed in Updating an Application to SQL Server Native Client from MDAC:

引用内容

If you use an ODBC or OLE DB call to start transactions, there is a difference in behavior between SQL Server Native Client and MDAC; transactions will begin immediately with SQL Server Native Client, but transactions will begin after the first database access using MDAC. This can affect the behavior of stored procedures and batches because SQL Server requires @@TRANCOUNT to be the same after a batch or stored procedure finishes execution as it was when the batch or stored procedure started. See Rollbacks and Commits in Stored Procedures and Triggers for more information.
With SQL Server Native Client, ITransactionLocal::BeginTransaction will cause a transaction to be started immediately. With MDAC the transaction start was delayed until the application executed a statement which required a transaction in implicit transaction mode. For more information, see SET IMPLICIT_TRANSACTIONS (Transact-SQL).

This statement was a bit difficult for me to understand. After further research and testing, verified by comparing the results of oledb calls and results of sql calls from SQL Server Management Studio, I came to following conclusion:

Within MDAC, ITransactionLocal::StartTransaction is basically an equivalent of SET IMPLICIT_TRANSACTIONS ON. This means the transaction will not start until one of the particular set of SQL statements executes. For the list of these statements, see SET IMPLICIT_TRANSACTIONS (Transact-SQL) . Do note Select is one of the statements that will cause the start of a transaction, if it is not in a transaction yet.
Within SQLNCLI, ITransactionLocal::StartTransaction is basically an equivalent of BEGIN TRANSACTION, which means an immediate start of a new transaction. But it can not be used to start a nested transaction. If a session has a transaction active, the repeated call of StartTransaction will be ignored.
With this difference in mind, the time of a transaction for a stored procedure call using SQLNCLI is possibly slightly longer than it was within MDAC, but the difference is likely negligible.
Unless absolutely required, it is not recommended to have your own transaction management stuff in a stored procedure. It will increase unnecessary burden and is prone to human errors. See Rollbacks and Commits in Stored Procedures and Triggers for more info.

According to API Implicit Transactions

引用内容

OLE DB does not have a method to set implicit transaction mode specifically.

    * Call the ITransactionLocal::StartTransaction method to start an explicit transaction.

    * When you then call either the ITransaction::Commit or ITransaction::Abort method with fRetaining set to TRUE, OLE DB completes the current transaction and goes into implicit transaction mode. The connection remains in implicit transaction mode as long as you set fRetaining on ITransaction::Commit or ITransaction::Abort to TRUE.

    * Call ITransaction::Commit or ITransaction::Abort with fRetaining set to FALSE to stop implicit transaction mode.

But my test is showing fRetaining MUST be false in ITransaction::Commit. This is proved by the MSDN library and actual testing (a TRUE value will result in error hr 0x8004d00f "An invalid combination of flags was specified").

If that's true, it means in SQLNCLI OleDB 2005/2008, there is no way to set the Implicit Transaction Mode. Which is not really a big deal since explicit transaction is probably better for clearer understanding anyway.

Reference:
SET IMPLICIT_TRANSACTIONS (Transact-SQL)
Updating an Application to SQL Server Native Client from MDAC
INFO: Understanding ADO Transactions with MS SQL Server
Nesting Transactions

[zt] Comparing Memory Allocation Methods

bluestrait@gmail.com(黄敏) — Tue,15 Dec 2009 18:26:59 +0800

Comparing Memory Allocation Methods

The following is a brief comparison of the various memory allocation methods:

Although the GlobalAlloc, LocalAlloc, and HeapAlloc functions ultimately allocate memory from the same heap, each provides a slightly different set of functionality. For example, HeapAlloc can be instructed to raise an exception if memory could not be allocated, a capability not available with LocalAlloc. LocalAlloc supports allocation of handles which permit the underlying memory to be moved by a reallocation without changing the handle value, a capability not available with HeapAlloc.

Starting with 32-bit Windows, GlobalAlloc and LocalAlloc are implemented as wrapper functions that call HeapAlloc using a handle to the process's default heap. Therefore, GlobalAlloc and LocalAlloc have greater overhead than HeapAlloc.

Because the different heap allocators provide distinctive functionality by using different mechanisms, you must free memory with the correct function. For example, memory allocated with HeapAlloc must be freed with HeapFree and not LocalFree or GlobalFree. Memory allocated with GlobalAlloc or LocalAlloc must be queried, validated, and released with the corresponding global or local function.

The VirtualAlloc function allows you to specify additional options for memory allocation. However, its allocations use a page granularity, so using VirtualAlloc can result in higher memory usage.

The malloc function has the disadvantage of being run-time dependent. The new operator has the disadvantage of being compiler dependent and language dependent.

The CoTaskMemAlloc function has the advantage of working well in either C, C++, or Visual Basic. It is also the only way to share memory in a COM-based application, since MIDL uses CoTaskMemAlloc and CoTaskMemFree to marshal memory.

error C2371: 'WCHAR' : redefinition

bluestrait@gmail.com(黄敏) — Mon,16 Nov 2009 17:00:11 +0800

I got this error today, when trying to switch an existing oledb project to use the new Microsoft SQL Server Native Client (referred to hereafter as SNAC) 10 instead of previous Microsoft Data Access Components (MDAC), to get the support of the new features introduced in MS SQL Server 2005 and 2008. For more info about this switching, see Microsoft SQL Server Native Client and Microsoft SQL Server 2008 Native Client.

As the fist step to introduce the definitions of SNAC 10, I added #include to the end of include list, immediately the compiler started to report following errors:

1>Compiling...
1>DBColumnSet.cpp
1>c:\devel\mhuangtfs\3rdparty\microsoft\sqlsdk\10.0\main\include\sqlncli.h(2736) : error C2371: 'WCHAR' : redefinition; different basic types
1> c:\program files\microsoft sdks\windows\v6.0a\include\winnt.h(289) : see declaration of 'WCHAR'

The relevant codes involved include:

sqlncli.h

#ifdef UNICODE
typedef wchar_t DBCHAR;
#else
typedef char DBCHAR;

#endif

winnt.h

#ifndef _MAC
typedef wchar_t WCHAR; // wc, 16-bit UNICODE character
#else
// some Macintosh compilers don't define wchar_t in a convenient location, or define it as a char
typedef unsigned short WCHAR; // wc, 16-bit UNICODE character
#endif

Various workarounds such as this one found from Internet were attempted but only to no avail. Eventually I found this article Using the SQL Server Native Client Header and Library Files . After I added in the macro definition, the compiler worked. Since I am using Oledb, my code is like below:

#define _SQLNCLI_OLEDB_
#include

Function Template

bluestrait@gmail.com(黄敏) — Sun,01 Nov 2009 18:35:14 +0800

Function template is compiled on demand. As function template is compiled on demand, the compiler needs a way to know the actual type to substitute the template parameter. It seems the only two ways are: either explicitly specify the type as in the example you have given, or make the type being used for one of the function parameters.

The chances are most of the time you do not even need to specify the type that you want the compiler to use - it will substitute them correctly, and complain if it finds an error. Each usage is compiled on demand, as and when the compiler encounters the references in the code.

However, you do need to specify the type if you don't reference the type as one of the parameter of the function prototype.

See following example:

template
void HelloTemplate()
{
T tTemp = _T("hello");
printf("%s", tTemp);
}

HelloTemplate();

The compiler will produce following error:

error C2783: 'void HelloTemplate(void)' : could not deduce template argument for 'T'

The reason appears to be the compiler is unable to determine what type is the T actually.
The code will work if you modify the line calling the function to be as below:

HelloTemplate();

Another case it will work is change the function prototype like below:

template
void HelloTemplate(T tParam)
{
printf("%s", tParam);
}

HelloTemplate(_T("Hello"));

Addressing MaxReceivedMessageSize issues

bluestrait@gmail.com(黄敏) — Sun,25 Oct 2009 19:18:18 +0800

You may come across this exception when calling a web service from a C# program:
“System.ServiceModel.CommunicationException : The maximum message size quota for incoming messages (65536) has been exceeded. To increase the quota, use the MaxReceivedMessageSize property on the appropriate binding element.”

This has something to do with the buffer size limit, the full explanation is in this blog:
http://craigrandall.net/archives/2009/04/maxreceivedmessagesize/.

Remember to update all three settings: maxBufferSize, maxBufferPoolSize, and maxReceivedMessageSize.

example app.config:

 程序代码

                                    receiveTimeout="00:10:00" sendTimeout="00:10:00" allowCookies="false"
                    bypassProxyOnLocal="false" hostNameComparisonMode="StrongWildcard"
                    maxBufferSize="1000000" maxBufferPoolSize="1000000" maxReceivedMessageSize="1000000"
                    messageEncoding="Text" textEncoding="utf-8" transferMode="Buffered"
                    useDefaultWebProxy="true">

Resource compiler: error RC2173

bluestrait@gmail.com(黄敏) — Tue,15 Sep 2009 22:22:29 +0800

This is a very weird problem I came across the other day.

I have a resource file whose first a few lines are as below:

// Microsoft Visual C++ generated resource script.
//
#include "IressRes.h"

#define APSTUDIO_READONLY_SYMBOLS
/////////////////////////////////////////////////////////////////////////////
//
// Generated from the TEXTINCLUDE 2 resource.
//
#include "afxres.h"
#include
/////////////////////////////////////////////////////////////////////////////
#undef APSTUDIO_READONLY_SYMBOLS

/////////////////////////////////////////////////////////////////////////////
// Neutral resources

#if !defined(AFX_RESOURCE_DLL) || defined(AFX_TARG_NEU)
#ifdef _WIN32
LANGUAGE LANG_NEUTRAL, SUBLANG_NEUTRAL
#pragma code_page(1252)
#endif //_WIN32

When I tried to compile, I got following errors from resource compiler:

1>.\Iress.rc(21) : error RC2173 : file .\Iress.rc is not a valid animated cursor/icon RIFF file

A search on Google on the reported error RC2173 returns nothing useful. Besides the error message is making no sense whatsoever, as there is not any animated curson/icon in the app at all.

After two hours fiddling around with the code, I found if I removed line 20 and line 23, the compiler would work. However, this does not make any sense. Almost driven crazy by this, I accidentally saved the file forcing to use Windows CR-LF as the line-ending (set in "Advanced Save Options" dialog) character, as I vaguely remembered for once VS asked me about that CR-LF thing at one stage, and only found out after that, the compiling succeeded.

After comparing the source code of the involved .rc file, it appears the problem was caused by the mismatching return character at one line of the resource file.

A bit backgournd about the new line character:
On Windows/DOS, a new line is represented by \r\n (i.e. 0x0D0A), while on Unix-like, it is a single \n (i.e. 0x0A), and \r (0x0D) on Mac.

In this case, for some reason, the line that caused the problem was ended with a single 0x0D, missing the paring 0x0A.

So forcing VS to save it using Windows CR-LF fix this newline character, and subsequently leads to the success of the resource compiler.

This is one of the weirdest problems I have ever seen with resource compiler. The error message and reported line position are not helpful at all, and it does not make sense for the compiler to fail just because of the missing 0x0A. I can only blame it on the lameness of the MS Resource Compiler.

Having said that, I suspected the missing of 0x0a in the first place is caused by some bugs of TFS.

The right permission to assigned to pjblog

bluestrait@gmail.com(黄敏) — Sun,30 Aug 2009 17:07:05 +0800

READ and Write permission must be assigned to the database folder otherwise checkuser.asp will complain unable to run some database operation (some update).

READ and Write permission must be assigned to Post and Cache folder. Otherwise, it will complain 3002: could not open file.

Removing unwanted source control binding

bluestrait@gmail.com(黄敏) — Tue,18 Aug 2009 21:38:33 +0800

Sometimes after you fiddled around with your projects within a solution by adding, removing, moving files around etc, you will find the newly added project may not be bound to source control any more, while its source files are still bound, when this happened, it may turn out that you can't find the menu "Add Selected Project to Source Control" any where.

When this happened, there are two ways to add the project back to source control:

1. Manually added the .vcproj file into source control as an new item.

2. or remove existing binding information to make the "add project to source control" appear again. Namely following lines appearing in a .vcproj file in Bold font must be removed:

 程序代码
    ProjectType="Visual C++"
    Version="9.00"
    Name="Price"
    ProjectGUID="{147EEF02-76DD-4668-9C99-51534E8A6A32}"
    RootNamespace="Price"
SccProjectName="SAK"
    SccAuxPath="SAK"
    SccLocalPath="SAK"
    SccProvider="SAK"
    Keyword="MFCDLLProj"
    TargetFrameworkVersion="131072"
    >

Also remove the associated .vspscc file.

After that, loading the solution will popup a warning dialog as you either remove the binding info for this project from the solution, or reuse the binding info from the solution. Choose to remove and then, your project will appear unbound from the source control. Now check the main menu and you should see the "add project to source control" again.

LNK2001: "int __cdecl _AfxInitManaged(void)"

bluestrait@gmail.com(黄敏) — Tue,18 Aug 2009 16:01:24 +0800

I got this error when I moved all source files (.h, .cpp, .rc) of a project from the original directory to another child directory, and then added them back to the project.

The problem is caused due to the fact that re-adding stdafx.cpp removed the preset configuration with respect to precompiled header. As explained in following thread:

http://social.msdn.microsoft.com/forums/en-US/vclanguage/thread/b3aa10fa-141b-4a03-934c-7e463f92b2a5/

引用内容

The fact that precompiled headers is solution wide and project wide property doesn't mean that the setting should be the same for all source files. Right click and choose properties for each and every one of your .cpp files, set the stdafx.cpp one to create and all other to use.

In detail, to fix the problem, from the solution pane, right-click the stdafx.cpp and open its properties, select All Configurations, then go to C/C++ -- Precompiled Headers page, and set Create/Use Precompiled Header property to be Create Precompiled Header (Yc).

书签： Scott Hanselman's ComputerZen.com

bluestrait@gmail.com(黄敏) — Mon,10 Aug 2009 15:51:23 +0800

Scott Hanselman's ComputerZen.com

including the very handy Rocky Scroll for VS 2005/2008.

Run Command Prompt Window as Administrator

bluestrait@gmail.com(黄敏) — Mon,27 Jul 2009 21:16:59 +0800

Lazy to type “CMD” or pressing any keyboard keys just to elevate the opened command prompt window with full administrator credentials and privileges mode?

Windows Vista
The following registry hack for Windows Vista will allow you to open an elevated command window by simply right click on Computer (formerly known as My Computer in Windows XP) icon at Start Menu or Desktop, and select Run Command Prompt as Administrator from the contextual menu.

To add and place a new menu item to right click context menu for Computer so that whenever you right click on the icon, a contextual menu item of Run Command Prompt as Administrator will be shown, and then you can straight away click on the entry to immediately open the elevated command prompt, unzip the enclosed .reg file from the attached zip and then import it into the registry database.

Windows 7
Manual way:
Search Bar
1. Click on Start and type CMD in search bar

2. Now press Ctrl+Shift+Enter

3. This should bring elevated command prompt with full Administrators rights.

Add Elevated Command Prompt Option in Right Click Context Menu of Folder in Windows 7
1. Downoad and unzip the enclosed reg file.

2. Right Click on AddElevatedCMDtoFolder.reg and select Merge from context menu.

3. Accept UAC prompt

4. Now Right click on any folder to very “Open Command Window Here (Administrator)” option is added or not.

References:
http://www.blogsdna.com/2226/windows-7-add-elevated-command-prompt-option-in-right-click-context-menu.htm

A on-the-fly localizing strategy

bluestrait@gmail.com(黄敏) — Tue,21 Jul 2009 21:56:54 +0800

http://stackoverflow.com/questions/751094/loading-caption-of-controls-from-resource-dll

I work on a large localized MFC project. Here is our strategy:

   1.      A dictionary of key -> localized string, specific to each language. There are a few ways to implement this, more later.

   2.      Control IDs or captions in the dialog resource are set to the key used to look up the translation

   3.      Create a base CDialog, CFormView, etc and in at init call ::EnumChildWindows. In the callback, look up the translation and replace the control's caption with the translation.

For your dictionary, you can go a few ways.

    *      If you want to rely on the built-in localized resource selection and string tables, you have to somehow match the control to the string ID. You can carefully ensure that the control ID matches the string ID, or you can ASCII-encode the ID in the caption and then use atoi to parse the int value.

    *      You can forgo the built-in localized string table deal and maintain your own string -> string dictionary for each language. This lets you set the caption to the non-localized string in the resource which makes layout easier (although you'll still need to test in all languages.) It will require you to do your own "dependency injection" to make sure you load up the right dictionary. You want to be able to release updated/additional languages without rebuilding the core binaries.

http://www.codeproject.com/KB/locale/MultiLangSupportMFCExtDLL.aspx

http://msdn.microsoft.com/en-us/library/8fkteez0%28VS.71%29.aspx
According to this,

Vista/Win 2008: registering an automation server

bluestrait@gmail.com(黄敏) — Sun,12 Jul 2009 17:53:39 +0800

On Vista or Vista-based OS (Windows 2008 server e.g.), when running an application from visual studio, which supports automation following line will produce errors:

m_server.UpdateRegistry(OAT_DISPATCH_OBJECT);
COleObjectFactory::UpdateRegistryAll();
AfxOleRegisterTypeLib(AfxGetInstanceHandle(), _tlid);

Error: failed setting key 'Phonebook.Document\CLSID' to value '{6309D913-88FF-4E2D-AC8B-E2FD580D527C}'.

As a result, the attempt to call this automation application from any other automation client app will fail with an error unable to create the object. For example, the test using following simple VBScript will produce error:

set wshell = CreateObject("Wscript.shell")

Dim obj
Set obj = CreateObject("Phonebook.Document")
obj.ShowWindow()

wscript.sleep 10000

Error 800A01AD

The problem appears to be Vista's annoying UAC (User Account Control) policy. This is preventing the running of debug application registering the server component in registry.

My workaround is to after building the application, run the .exe from Windows explorer, using "Run as administrator" from righ-click menu.

Login failed for user xxx to SQL Server 2008

bluestrait@gmail.com(黄敏) — Tue,07 Jul 2009 21:19:24 +0800

On Windows 2008 Server (64bit), Tried to connect to the database using SQL server Management Studio 2008 using Windows Authentication, receiving an error:

Cannot connect to (local).

===================================

Login failed for user 'DEVEL\mhuang'. (.Net SqlClient Data Provider)

------------------------------
For help, click: http://go.microsoft.com/fwlink?ProdName=Microsoft+SQL+Server&EvtSrc=MSSQLServer&EvtID=18456&LinkId=20476

------------------------------
Server Name: (local)
Error Number: 18456
Severity: 14
State: 1
Line Number: 65536

Found the trick here to fix this problem is to I should have started SQL Management Studio as Administrator. This issue only occurs on Vista/2008 Server R2.

[2010-6-29 Update] Similar issue is occurring again after I switched back from SQL authentication to Windows Authentication and this time, the trick above does not work.

The cause is because the windows user account was not set up as a login in sql server. The detailed solution is in http://docs.google.com/document/pub?id=1UO1v69QcWdgwifXwfB6EGZg_YbiKma5cyx4XTGZuM8A.

Script error in VC++ wizards after installing

bluestrait@gmail.com(黄敏) — Mon,29 Jun 2009 16:06:43 +0800

You might have script errors in VC++ wizards after installing IE8, for example, trying to add a member variable using class wizard, or when you double click inside a dialog normally we should taken to corresponding file but instead we get a script error…

Script error

Pretty strange isn’t it. Well as per VC++ team blogs this bug due to IE8 installation. Reason for this behavior is as explained below (quoted from VC++ team blogs…)

The VC++ Wizards Engine implements the IInternetSecurityManager interface. In this implementation it allows or disallows specific actions under certain policies that Internet Explorer queries it about. In IE8 a custom Security Manager now also gets queried about the URLACTION_ACTIVEX_OVERRIDE_REPURPOSEDETECTION policy which IE previously did not delegate to custom security managers when the engine wasn’t running in the iexplore.exe process. The IE engine then fails this action because we don’t have a policy entry for it in the custom zone for VC++ Wizards. We are still investigating whether this change in IE8 is by design and will possibly be addressing it by a fix in either the Wizard or IE components depending on the outcome.

And according to this blog, following fix should work….

Please follow the following steps:

Open regedit (on a 64-bit OS, open the 32-bit regedit)
Under “HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet
Settings\Zones”, create a new key called 1000 (if it isn’t already there)
Under 1000, create a DWORD entry with:
- o Name = 1207
- o Type = REG_DWORD
- o Data = 0×000000

Just now I had this problem hence I had to use this fix and it works.

How to remote root login over ssh

bluestrait@gmail.com(黄敏) — Wed,17 Jun 2009 19:24:16 +0800

In /etc/ssh/sshd_config there is a line as below:

PermitRootLogin no

Change "no" to "yes" then restart sshd service to do the trick.

C++ FAQ LITE

bluestrait@gmail.com(黄敏) — Mon,11 May 2009 21:11:28 +0800

http://www.parashift.com/c++-faq-lite/

Use regular expressions to remove blank lines

bluestrait@gmail.com(黄敏) — Mon,27 Apr 2009 13:47:59 +0800

Remove blank lines

A question which is frequently asked is "I have a lot of blank lines in my file and I don't want to go through and manually delete them. Is there an easier way to do this?"

The answer: YES!

In versions of UltraEdit that support Perl-compatible regular expressions, you can use the following Perl-compatible regular expression in DOS/Unix/Mac formatted-files. You can enable Perl-compatible regular expressions under Advanced -> Configuration -> Search -> Regular Expression engine.

Replace: "^\r?\n?$" (without the quotes)
With "" (without the quotes - i.e. nothing).

In previous versions of UltraEdit, to delete blank lines you can use a regular expression replace as follows: Find What: "^p$" (without the quotes)
Replace With: "" (without the quotes - i.e. nothing).

Removing Blank lines

Step 1: Determine the file(s) that contain blank lines

Step 2: Open the Replace dialog

You may open the replace dialog using the shortcut (CTRL + R) or by going to the search menu and clicking on "Replace".

To delete blank lines you can use a regular expression replace as follows:

Find What: "^p$" (without the quotes)
Replace With: "" (without the quotes - i.e. nothing).

Note: You MUST select "Regular Expressions." Also, "UltraEdit style expressions" must be selected under Advanced -> Configuration -> Search -> Regular Expression Engine.

Depending on your circumstances, the "Replace Where" option you select may vary.

Step 3: Your blank lines are gone!

Please note, you may have to run the "Replace All" multiple times to ensure all blank lines are removed. Because of the way the regular expression works, the replace will not replace all blank lines that are grouped together.

NOTE: From my experience, I have found "^\r?\n?" works in TextPad, but not "^\r?\n?$". It seems TextPad does not understand "$".

[zt] Which process listens to this port (in Solari

bluestrait@gmail.com(黄敏) — Tue,31 Mar 2009 16:31:40 +0800

If you have 'lsof' installed, it is very easy to find out which process is listening at certain port. However, in Solaris, it may not be so obvious. This is the question my friend asked me and I wrote a short script (listen.sh) for him. Basically it runs 'pfile' over all the processes listed in /proc (except process 0, which is the sched)

#! /bin/sh

if [ $# -ne 1 ]; then
  echo "Usage: $0 "
  exit 1
fi
listen=$1

PATH=/usr/bin:/bin
export PATH

#
# skip process 0
#
cd /proc
for i in [1-9][0-9]*
do
  pfiles $i | nawk -v listen=$listen '
    BEGIN {
      found=0
    }
    NR==1 {
      process=$0
    }
    /sockname/ && $NF == listen {
      getline
      if ( ! /peername/ ) {
        found=1
        exit
      }
    }
    END {
      if ( found == 1 ) {
        printf("%s\n",process)
      }
    }'
done

For example:
# ./listen.sh 22
29626: /usr/lib/ssh/sshd
#

this script is great - just what I needed! I adopted it in the following way such that it won't output error messages of processes which the uses who runs this script does not have access to, modify the line
pfiles $i 2> /dev/null | nawk ...

such that strerr will be redirected to /dev/null

Replacing with a new line character

bluestrait@gmail.com(黄敏) — Thu,05 Feb 2009 15:39:06 +0800

In vi, hit ctrl+v then tap return (enter). This pastes in a ^M which is actually a newline character.

It will look like below for example:

:%s /^@/^M/g

This is to replace some strange control character (Ctrl-@) with a new line character.

How to set the output of a command to a variable

bluestrait@gmail.com(黄敏) — Fri,28 Nov 2008 20:51:30 +0800

to set the output of a command to a variable in DOS extension:

 程序代码
FOR /F "tokens=*" %%R IN ('PROGRAM.EXE /SWITCH') DO SET VAR=%%R

Remove ^M and other control characters in vi

bluestrait@gmail.com(黄敏) — Tue,11 Nov 2008 09:03:48 +0800

To remove the ^M characters at the end of all lines in vi, use:

:%s/^V^M//g

The ^v is a CONTROL-V character and ^m is a CONTROL-M. When you type this, it will look like this:

:%s/^M//g

In UNIX, you can escape a control character by preceeding it with a CONTROL-V. The :%s is a basic search and replace command in vi. It tells vi to replace the regular expression between the first and second slashes (^M) with the text between the second and third slashes (nothing in this case). The g at the end directs vi to search and replace globally (all occurrences).Remove ^M characters at end of lines in vi

How to link a directory outside htdocs to Apache

bluestrait@gmail.com(黄敏) — Tue,19 Aug 2008 09:56:38 +0800

Use alias directive, for example:

 程序代码
    
Alias /sharedfiles/ "D:\Personal\Video\2008BeijingOpening/"     

Note the trailing slash is a backward slash instead of forward slash. This is important.

After that, you should be able to access the content in that directory by using a url like below:
http://:/shardfiles/

Timezone issue with PJBlog

bluestrait@gmail.com(黄敏) — Wed,23 Jul 2008 12:08:49 +0800

It is quite common that the web server is hosted on a different physical location from where you run it, sometimes even on a different continent. This expose a problem when your server side code is getting system time for timestamps, which later got displayed on the web page and confuse you.

I believe that means your code should be flexible enough to handle this sort of situation, at least customizable.

PJBlog, the server hosting this blog, unfortunately does not seem to be capable to handle this issue. It uses now() function widely to get timestamp from the server and set it to the posts, articles, comments, etc. For me, it is a problem as my server is located in US, while I am working from Australia.

I checked the source code of PJBlog, and it seems there is no any place that you can work around it, except to modify the code where now() function is called, so that a correct timezone difference (17 hours in my case) will be added to return value from now() before it got used by the application.

The attempt to make a "global" function so that it can be included in each asp script that has now() referenced failed too, due to some recursive include issue among these pages. Shame.

.NET framwork issue with IIS on Windows XP

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 18:10:39 +0800

Had following issues:
Server Application Unavailable

on asp.net page, but asp seems fine. This is indicating .net framework issue.
Event log is showing:

Failed to execute request because the App-Domain could not be created. Error: 0x80131902

also stopping iis has failed, and event viewer is showing:

Failed while restarting W3SVC. Error code: 0x8007041d

Fixes:
1. use previous blog to create a new site on IIS, set the root folder.
2. right-click the root folder and set the folder to be web shared as "/" for the new site, in the properties window of the folder.
3. uninstalled .net framework 1.0
4. ran

 程序代码
C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727> aspnet_regiis.exe -i

to install 2.0 framework only
5. ran

 程序代码
C:\WINDOWS\system32>iisreset

to reset/restart IIS.
6. after that, in IIS service panel, stop default site, and start the new site.

[zt] Multiple websites using IIS 5.1 on Windows XP

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 17:10:59 +0800

Ever wondered how you can set up multiple websites in Windows XP (i.e. Proffesional)?
Did you wonder at least if it possible?

Well, it is. Microsoft has it’s own way of hiding things…from novice users, so that the system will not crash down… Hmm, just a small thought: is this how you should protect something? By not letting people get their hands on all of it?

Another small thought: Microsoft is more like…building this car, lots and lots of advertising to gather a big percentage of the market, you buy it, looks very nice with all those visual styles…iuhuu! :) …then, when something goes wrong with the car..and you just want to take a peak….uppps!… The hood is locked.
But still, Microsoft is Microsoft and few wackos can deny this…

Back to our stuff.
Inside the INETPUB folder that Windows XP creates when installing IIS, there’s a folder called ADMINSCRIPTS…
Open a Command Prompt in that folder.
By typing

 程序代码
adsutil.vbs enum w3svc /p

you will get a list of websites created. Basically only the Default Website (W3SVC/1)

To create a new one, type

 程序代码
adsutil.vbs create_vserv W3SVC/2

(the last number can be anything; it is the index of the new website)

 程序代码
adsutil.vbs copy W3SVC/1 W3SVC/2

and go to the Internet Information Services/Computer Management and rename this new website, give it a new root folder and set it up however you like.

To delete one, type

 程序代码
adsutil.vbs delete W3SVC/2

What you have to remember is that in Windows XP, you can only have one, and only one website running. That means, if you want to start the new site just created, you have to stop the default one first.

Of course, there is a simpler way – free software:
http://www.firstserved.net/services/iisadmin.php
http://jetstat.com/iisadmin/

You can also try http://www.hairy-spider.com/multisite.aspx, which is an ISAPI filter that can associate different domains to 127.0.0.1 and differenciates requests through that. It doesn’t deliver website alternative configuration though.

Opening a DOS Prompt from Windows File Explorer

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 09:17:12 +0800

Windows 2000, Windows XP
This trick works on both Window 2k and XP with minor differences between
dialogs.

You can use the Explorer to easily open up a DOS window to a specific
directory.

  1. From Explorer Option Menu select File Types
  2. Highlight the object for Folder - NOT File Folder. If you can't
find it, click "new" to create a new file extension, type in "*" as the
file extension, and click "Advanced"  so that you can see Associated
file type: Select "Folder" as the file type.
  3. Click the Edit/Advanced button
  4. From the 'Edit File Type' panel click the 'New' button
  5. Action is '&Goto DOS Prompt'
  6. Application used is cmd.exe /k pushd %L
  7. Select OK, then Close and Close.
  8. Now from Explorer secondary (right) click the folder you want to
go to in DOS
  9. From the menu select 'Goto DOS Prompt' (or hit G) and you are in
that directory with a DOS windows

Windows Vista
Here comes Windows Vista and ads this very useful shortcut to the DEFAULT installation inside Windows Explorer's context menu!

   1.      Open a Windows Explorer windows, browse to the required folder.

   2.      Right-click that folder in the right pane of the Windows Explorer window. Note that you do NOT have the "Open Command Prompt Here" option.

   3.      Now, hold the SHIFT key while you right-click the folder. Behold!

Note: As in most cases, Microsoft has only gone part of the way with this cool feature. It's silly but this context menu add-on in only available when you right-click on the folder in the right pane, and not in the left pane or in a My Computer window...

A cool feature of this context menu add-on is that in case of a network share it will automatically and silently map a drive to that share, and disconnect the mapped drive when the Command Prompt closes.

Recording output of series of commmands on Linux

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 09:03:36 +0800

The basic idea is to make use of script command.

Usage:

 程序代码
$ script 
 
$ exit

After that, logfilename should have all the output on the screen recorded.

Example:

[oracle@mhuang-au2 ~]$ script templog
Script started, file is templog
[oracle@mhuang-au2 ~]$ pwd
/home/oracle
[oracle@mhuang-au2 ~]$ ls -ltr
total 28248
-rw-r--r--   1 oracle oinstall  4869710 Dec 28  2006 php-4.3.9.tar.gz
drwxr-xr-x  16 oracle oinstall     4096 Dec 28  2006 php-4.3.9
drwxr-xr-x   3 oracle oinstall     4096 Jan  3  2007 userlog
drwxr-xr-x   2 oracle oinstall     4096 Jan  4  2007 Desktop
drwx------   5 oracle oinstall     4096 Jan 12  2007 oraInventory
drwxrwx---   2 oracle oinstall     4096 Jan 12  2007 oraDev
drwxrwx---   4 oracle oinstall     4096 Mar  9  2007 oradata
-rw-rw----   1 oracle oinstall 23840481 Apr  3  2007 installation.report.txt
drwxrwx---   3 oracle oinstall     4096 Apr  3  2007 oraHome_BI
drwx------   2 oracle oinstall     4096 Apr 30  2007 wlt_testlbr
drwxrwx---   3 oracle oinstall     4096 May  1  2007 oraHome_6
drwxrwx---   2 oracle oinstall     4096 May  1  2007 oraHome_7
drwx------   2 oracle oinstall     4096 May  2  2007 testwallet2
drwxrwx---   3 oracle oinstall     4096 May 15  2007 oradata2
drwx------  81 oracle oinstall     4096 Jun  5  2007 oraHome_5
drwx------   2 oracle oinstall     4096 Jun 21  2007 testwallet3
drwx------  74 oracle oinstall     4096 Jun 22  2007 oraHome_TestInfra
-rwxr--r--   1 oracle oinstall      113 Aug  1  2007 setinfra1env
-rwxr--r--   1 oracle oinstall      129 Aug  1  2007 setinfra2env
drwxrwx---   3 oracle oinstall     4096 Aug  9  2007 product
drwxr-xr-x   3 oracle oinstall     4096 Aug 10  2007 testweb
drwxrwxr-x   8 oracle oinstall     4096 Aug 14  2007 installcds
drwxrwx---   2 oracle oinstall     4096 Sep  5  2007 oraHome_8
drwxr-xr-x   2 oracle oinstall     4096 Sep 10  2007 testweb4
drwxr-x---  56 oracle oinstall     4096 Sep 13  2007 soa_10131_basic
drwxr-x---  55 oracle oinstall     4096 Sep 20  2007 oraHome_soa10131_j2ee
drwxr-xr-x   4 oracle oinstall     4096 Oct  3  2007 patches
-rw-r--r--   1 oracle oinstall      393 Oct 16  2007 README.MIN
drwxr-xr-x   3 oracle oinstall     4096 Oct 25  2007 loadzone
drwxr-xr-x   5 oracle oinstall     4096 Jan 21 10:40 testweb2
drwxrwx---   2 oracle oinstall     4096 Feb  5 07:11 oraHome_9
drwx------  76 oracle oinstall     4096 Apr 23 08:40 oraHome_1
-rw-r--r--   1 root   root         1005 Apr 23 11:27 profile.bak
drwxr-xr-x   5 oracle oinstall     4096 Apr 30 09:20 JAT
-rw-r--r--   1 oracle oinstall      320 May  1 07:34 jsptest.jsp
drwxr-x---  42 oracle oinstall     4096 May 26 11:44 oraHome_soa10131_http
drwxr-xr-x   2 oracle oinstall     4096 May 29 10:46 lab
drwx------  21 oracle oinstall     4096 Jun 10 11:57 mechan
drwx------  61 oracle oinstall     4096 Jun 11 12:13 mid101202
drwxr-xr-x   4 oracle oinstall     4096 Jul  4 06:59 rdatool
drwx------  62 oracle oinstall     4096 Jul 15 07:47 oraHome_3
drwx------  86 oracle oinstall     4096 Jul 15 13:29 oraHome_2
-rw-r--r--   1 oracle oinstall     7705 Jul 21 07:31 temp.log
-rw-r--r--   1 oracle oinstall        0 Jul 21 07:32 templog
[oracle@mhuang-au2 ~]$ exit
exit
Script done, file is templog

[oracle@mhuang-au2 ~]$ cat templog
Script started on Mon 21 Jul 2008 07:32:56 AM EST
[oracle@mhuang-au2 ~]$ pwd
/home/oracle
[oracle@mhuang-au2 ~]$ ls -ltr
total 28248
-rw-r--r--   1 oracle oinstall  4869710 Dec 28  2006 php-4.3.9.tar.gz
drwxr-xr-x  16 oracle oinstall     4096 Dec 28  2006 php-4.3.9
drwxr-xr-x   3 oracle oinstall     4096 Jan  3  2007 userlog
drwxr-xr-x   2 oracle oinstall     4096 Jan  4  2007 Desktop
drwx------   5 oracle oinstall     4096 Jan 12  2007 oraInventory
drwxrwx---   2 oracle oinstall     4096 Jan 12  2007 oraDev
drwxrwx---   4 oracle oinstall     4096 Mar  9  2007 oradata
-rw-rw----   1 oracle oinstall 23840481 Apr  3  2007 installation.report.txt
drwxrwx---   3 oracle oinstall     4096 Apr  3  2007 oraHome_BI
drwx------   2 oracle oinstall     4096 Apr 30  2007 wlt_testlbr
drwxrwx---   3 oracle oinstall     4096 May  1  2007 oraHome_6
drwxrwx---   2 oracle oinstall     4096 May  1  2007 oraHome_7
drwx------   2 oracle oinstall     4096 May  2  2007 testwallet2
drwxrwx---   3 oracle oinstall     4096 May 15  2007 oradata2
drwx------  81 oracle oinstall     4096 Jun  5  2007 oraHome_5
drwx------   2 oracle oinstall     4096 Jun 21  2007 testwallet3
drwx------  74 oracle oinstall     4096 Jun 22  2007 oraHome_TestInfra
-rwxr--r--   1 oracle oinstall      113 Aug  1  2007 setinfra1env
-rwxr--r--   1 oracle oinstall      129 Aug  1  2007 setinfra2env
drwxrwx---   3 oracle oinstall     4096 Aug  9  2007 product
drwxr-xr-x   3 oracle oinstall     4096 Aug 10  2007 testweb
drwxrwxr-x   8 oracle oinstall     4096 Aug 14  2007 installcds
drwxrwx---   2 oracle oinstall     4096 Sep  5  2007 oraHome_8
drwxr-xr-x   2 oracle oinstall     4096 Sep 10  2007 testweb4
drwxr-x---  56 oracle oinstall     4096 Sep 13  2007 soa_10131_basic
drwxr-x---  55 oracle oinstall     4096 Sep 20  2007 oraHome_soa10131_j2ee
drwxr-xr-x   4 oracle oinstall     4096 Oct  3  2007 patches
-rw-r--r--   1 oracle oinstall      393 Oct 16  2007 README.MIN
drwxr-xr-x   3 oracle oinstall     4096 Oct 25  2007 loadzone
drwxr-xr-x   5 oracle oinstall     4096 Jan 21 10:40 testweb2
drwxrwx---   2 oracle oinstall     4096 Feb  5 07:11 oraHome_9
drwx------  76 oracle oinstall     4096 Apr 23 08:40 oraHome_1
-rw-r--r--   1 root   root         1005 Apr 23 11:27 profile.bak
drwxr-xr-x   5 oracle oinstall     4096 Apr 30 09:20 JAT
-rw-r--r--   1 oracle oinstall      320 May  1 07:34 jsptest.jsp
drwxr-x---  42 oracle oinstall     4096 May 26 11:44 oraHome_soa10131_http
drwxr-xr-x   2 oracle oinstall     4096 May 29 10:46 lab
drwx------  21 oracle oinstall     4096 Jun 10 11:57 mechan
drwx------  61 oracle oinstall     4096 Jun 11 12:13 mid101202
drwxr-xr-x   4 oracle oinstall     4096 Jul  4 06:59 rdatool
drwx------  62 oracle oinstall     4096 Jul 15 07:47 oraHome_3
drwx------  86 oracle oinstall     4096 Jul 15 13:29 oraHome_2
-rw-r--r--   1 oracle oinstall     7705 Jul 21 07:31 temp.log
-rw-r--r--   1 oracle oinstall        0 Jul 21 07:32 templog
[oracle@mhuang-au2 ~]$ exit
exit

Script done on Mon 21 Jul 2008 07:33:01 AM EST

VIM Tricks

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 07:35:21 +0800

Movement commands

G
    Move to the end of the file.

~CTRL-g
   Display file summary information.

gg
    Move to the beginning of the file.

~CTRL-e
    Expose another line at the top. Useful for scrolling the screen without moving your cursor from its position.

~CTRL-y
    Expose another line at the bottom. Useful for scrolling the screen without moving your cursor from its position.

}
    Move forward a paragraph. Useful for moving through documents which have blank lines separating paragraphs or blocks. An easy way to distinguish between this and { is to remember it points in the direction you wish to go.

{
    Move backward a paragraph. Useful for moving through documents which have blank lines separating paragraphs or blocks.

)
    Move forward a sentence. An easy way to distinguish between this and ( is to remember it points in the direction you wish to go

(
    Move backward a sentence.

:number
    Where number is a decimal number indicating the line number you wish to jump/move to.

'0
    Vim creates a "bookmark" each time you exit Vim. The last "bookmark" is '0. So if you have a file you haven't edited in a long time and can't remember where you left off simply type this command and it will take you to the place you last edited. You can find out what marks are set by typing :marks.

Search Commands

''*''
    This is a search command which will find the next occurance (forward) of the word the cursor is currently on.

n
    Find the (n)ext occurance of the previous search. It will use the same direction (i.e. forward vs. backwards) of the previous search as well.

Simple editing/combination commands

Deleting/selecting parts of a line

dfx
    Delete forward until you find x (inclusive). If x is found, vim will delete everything from the current cursor position to and including x.

    This one and its variations dFx, dtx, and dTx are very useful when editing mark-up languages like html and sgml. It is especially useful in deleting and changing quoted attributes within mark-up tags.

    This is a combination of the (d)elete command and the (f)ind motion command. You may also substitute other commands for the delete portion of that command, such as visual by character, or yank.

dFx
    Delete backward until you find x (inclusive). If x is found, vim will delete everything from the current cursor position to and including x.

dtx
    Delete forward until you find x (exclusive). If x is found, vim will delete everything from the current cursor position to, but not including x.

dTx
    Delete backward until you find x (exclusive). If x is found, vim will delete everything from the current cursor position to, but not including x.

Deleting/selecting spanning multiple lines

d/x
    Delete forward until you find x (exclusive). If x is found, vim will delete everything from the current cursor position to, but not including x, spanning multiple lines if necessary.

    x doesn't have to be one character. It's most likely to be a word or several characters. If you're editing your email, you could use this to delete all text up to your signature by executing d/--.

    This is a combination of the (d)elete command and the (/) search forward motion command. You may also substitute other commands for the delete portion of that command, such as visual by character, Visual by line, or yank.

d?x
    Delete backward until you find x (exclusive). If x is found, vim will delete everything from the current cursor position to, but not including x, spanning multiple lines if necessary.

Checking the processes using a particular TCP port

bluestrait@gmail.com(黄敏) — Mon,21 Jul 2008 07:15:14 +0800

This is the command that can be used to display the processes using a particular TCP port:

lsof -i TCP:

Note: lsof typically is located in /usr/sbin and unfortunately it may not be in default PATH.

Example:

 程序代码

[oracle@mhuang-au2 bin]$ /usr/sbin/lsof -i :7779
COMMAND   PID   USER   FD   TYPE   DEVICE SIZE NODE NAME
httpd   25976 oracle   22u  IPv4 70480520       TCP *:7779 (LISTEN)
httpd   25997 oracle   22u  IPv4 70480520       TCP *:7779 (LISTEN)
httpd   26000 oracle   22u  IPv4 70480520       TCP *:7779 (LISTEN)

After that, we can query process information using the PID found.