PC-lint嵌入VS2005 总结
首先得到vs2005.lnt, co-msc80.lnt, env-vc8.lnt三个文件，具体内容看下文
步骤
1. 安装pc-lint或拷贝已安装的文件到已指定目录
2.VS2005->tools->external tools
3. add pclint(名字不定)
4. commond,选择pclint的exe文件
5.arguments
-i“vs2005.lnt文件目录” -u vs2005.lnt "$(ItemDir)$(ItemFileName)$(ItemExt)"（-i与后面内容没有空格）
其他用刀的lnt文件需要同vs2005.lnt在一个目录下
6 使用PC-Lint时可能会遇到很多库的头文件的错误，特别是在VS中。这时可以通过设置PC-LINT.exe的参数屏蔽库文件的错误。一般为PC-LINT.exe -wlib(x)
-wlib(0) 不生成任何库信息
-wlib(1) 只生成错误信息（当处理库的源代码时）
-wlib(2) 生成错误和告警信息
-wlib(3) 生成错误、告警和其它信息（这是默认设置）
-wlib(4) 产生所有信息
另外，如果你测试时不想看一下info类的信息，可以在命令中增加-wx来屏蔽不需要显示的信息：
-w0 不产生信息（除了遇到致命的错误）
-w1 只生成错误信息 -- 没有告警信息和其它提示信息
-w2 只有错误和告警信息
-w3 生成错误、告警和其它提示信息（这是默认设置）
-w4 生成所有信息
一般情况下我先使用-w1再用-w2，最后用-w3
vs2005.lnt内容，注意-i和后面内容之间没有空格
au-sm.lnt
co-msc80.lnt
lib-mfc.lnt
lib-stl.lnt
lib-w32.lnt
lib-wnt.lnt
lib-atl.lnt
options.lnt -si4 -sp4
-i"E:/Program Files/Microsoft Visual Studio 8/VC/include"
-i"E:/Program Files/Microsoft Visual Studio 8/VC/atlmfc/include"
-i"E:/Program Files/Microsoft Visual Studio 8/VC/PlatformSDK/include"
-i"E:/Program Files/Microsoft Visual Studio 8/SDK/v2.0/include"
-i"F:/exercises/set/set"
co-msc80.lnt
// co-msc80.lnt
// Compiler Options for Microsoft C Version 14.00
// (32 bit version)
// This file contains options to allow PC-lint to process source
// files for your compiler. It is used as follows:
//
// lint co-msc80.lnt source-file(s)
//
-cmsc
-si4 // integers are 4 bytes
-sp4 // pointers are 4 bytes too.
// We now support __declspec directly so that the following
// option is now commented out. If trouble ensues you can
// once again disable __declspec through this option.
// -d__declspec()= // ignore this construct
-d_declspec=__declspec // the single '_' version is occasionally used
// while processing compiler (library) header files ...
-wlib(1) // sets the warning level within library headers to 1
// (no warnings, just syntax errors). Comment out if you
// are actually linting library headers. This
// option makes obsolete options of the form -elib(axxx) where
// xxx >= 400 which may be retained for historical reasons.
-elib(19) // useless declarations (lone semicolons)
-elib(123) // function-like macro name used as non macro
-elib(652) // suppress message about #define of earlier declared symbols
-elib(762) // suppress message about multiple identical declarations and
-elib(760) // suppress message about multiple identical macro defs
-elib(514) // allow #if <boolean> | <boolean>
-elib(553) // undefined preprocessor variables assumed 0
-elib(726) // extraneous comma in enum definition
// SREGS, WORDREGS, BYTEREGS are defined in both bios.h and dos.h
// and accordingly you MAY get type differences based on 'origin'.
// If so, use the following options:
// -etd(origin)
// -elib(770)
-format=%(%f(%l)/s:/s%)%t/s%n:/s%m
// error format similar to MSC
// Note that %c can also be used to specify column
-e46 // allows bit-fields to be other than int or unsigned
+fan // allow anonymous unions
+fdi // Use directory of the including file
+fbo // enable the bool type
-fdh // do not append a .h to header names
-esym(123,min,max) // allows users to use min, max as variables
+rw(__inline) // activate the __inline keyword
+ppw(import) // activate #import
-d_inline=__inline // _inline is the same as __inline
-sld10 // sizeof(long double) is 10.
-function(exit,_exit) // _exit() is like exit()
-function(exit,_assert) // _assert() is like exit()
-emacro(506,assert) // don't warn about constant value Boolean
-emacro(734,putc) // don't complain about items being too large.
-emacro(415,_FP_SEG) // access of out-of-bounds pointer
-emacro(740,FP_SEG,FP_OFF) // unusual casts
-emacro((???),va_arg) // the va_arg() macro can yield 415, 416, 661, 662
// 796 and 797 (out-of-bounds errors).
-emacro((???),va_start) // the same applies to va_start
-emacro(413,offsetof) // use of NULL pointer creates a stir
-emacro(545,offsetof) // addressing an array member is OK
-e793 // inhibit 'ANSI limit reached' --
// limits are impractically low with MSVC headers
-esym(628,eof) // failure to provide argument information for eof()
-esym(773,L_tmpnam) // defined with an unparenthesized '+'
// The following functions exhibit variable return modes.
// That is, they may equally-usefully be called for a value
// as called just for their effects. Accordingly we inhibit
// Warning 534 for these functions.
// Feel free to add to or subtract from this list.
-esym(534,close,creat,fclose,fflush,_flsbuf,fprintf,fputc)
-esym(534,fputs,fscanf,fseek,fwrite,lseek,memcpy,memmove,memset)
-esym(534,printf,puts,scanf,sprintf,sscanf,strcat,strcpy)
-esym(534,strncat,strncpy,unlink,write)
// These are the wide char variants of printf-scanf family
-wprintf( 1, wprintf )
-wprintf( 2, fwprintf, swprintf )
-wscanf( 1, wscanf )
-wscanf( 2, fwscanf, swscanf )
// The following options are used to adjust our function mimicry to
// the actual library as provided by MS.
-function( wcstombs(1) ) // remove the check for a NULL first arg to wcstombs()
// The following options are required by most compilers to
// noiselessly process iostream.h
-elib(1717) //empty prototypes
-elib(522) //function return value ignored
-elib(1053) //prototypes cannot be distinguished
-elib(1721) //unusual operator =() declaration
-elib(1720) //assignment op has non-const parameter
-elib(655) // bitwise operator combining compatible enum's
-elib(641) // converting enum's to int
-elib(537) // repeated include file (ios.h)
-elib(1511) // member (rdbuf) hides nonvirtual member
-elib(1712) // default constructor not defined for class
-elib(1736) // redundant access specifier d_NATIVE_WCHAR_T_DEFINED
-esym(1702,operator<<,operator>>) // both a member and an ordinary function
// These functions return things that are frequently ignored.
-esym(534,*operator<<,*operator>>)
// The following additional options seem to be needed.
-elib(506) // constant value Boolean
-elib(620) // el or one? (some constants end in 'l' not 'L')
-elib(648) // overflow in computing constant (3<<16)
-elib(659) // nothing fold_NATIVE_WCHAR_T_DEFINEDlows '}' on some line
-elib(723) // suspicious use of '='
-elib(747) //significant prototype coercion
-elib(740) //unusual pointer casts
-elib(1007) // virtual functions within extern "C" block
-elib(1029) //default argument repeated -- can't dist. char, signed char
-elib(1055) //call to rdbuf() questioned?
-elib(1504) // apparently useless structs
-elib(1708,1709) // minor C/C++ declaration conflict
-elib(1707) // operator new declared w/o 'static'
-elib(1722) // assignment op does not return reference
-elib(149) // default argument found in C code.
-elib(578) // declaration of time hides delaration of global time().
-elib(761) // two equivalent typedef declarations
-elib(1065) // same name declared as "C" and not "C"
-elib(1066) // same name declared as "C" and not "C"
-elib(1704) // constructor with private access declaration
-elib(1735) // default parameter within virtual function
-elib(773) // macros that look like unparenthesized expressions
-elib(806) // 1-bit bitfields typed int
-elib(1501) // 0-length data members
-elib(1510) // base class has no destructor
-elib(1516) // data member hides inherited member
-elib(1509) // base class destructor is not virtual
// Special Notice: You may be receiving mysterious 1058 errors
// when you use "iomanipulator"s. For example:
// cout << setw(4) << 4;
// results in Error 1058 (assigning a const to a ref) because the
// manipulator setw returns a non-lvalue which is assigned to a reference.
// This reflects an oversight in the Microsoft header file: iomanip.h
// Therein you may change the declaration:
// friend ostream& operator<<(iostream& s, IOMANIP(T) & sm) { ...
// to:
// friend ostream& operator<<(iostream& s, const IOMANIP(T) & sm) { ...
// to reflect the fact that sm is not modified by this function.
-d_INTEGRAL_MAX_BITS=64 // denotes support for __int64
+fll // enable long long
-d__STDC__=0 // Starting with 2.0, the tests within Microsoft header
// files for __STDC__ are of the form #if rather than #ifdef
-dWIN32 // this is necessary because without WIN32 being defined
-d_WIN32 // a very gross syntax error is exposed in the objbase.h
// ... definition of function operator ==(). Alternatively,
// ... repair the function definition.
-dDBG=0 // There is a #if test made for this variable.
/* The option -d_NEW_ previously appearing in this spot is no longer
recommended as it nullifies the <new> header and because of this
prevents definition of 'bad_alloc' among other things.
*/
/* DLL's or Multithreads? Enable the following:
-d_AFXDLL // making a DLL
-d_DLL // ditto
-d_MT // multi-thread
*/
/* At one time we indicated that wchar_t was predefined by using
the option +fwc but then because to suppress the Microsoft headers
from redefining it we used the option -d_WCHAR_T_DEFINED
But this in turned created a rare problem so now we just leave it
undefined.
*/
// In the following option we define __uuidof() and suppress
// Errors 50 and 69 and 1924 in exprs. containing same
-d"__uuidof()= /*lint --e(50,69,1924) */ (_GUID)0"
-esym(123,FD_SET) // defined as macro and as typedef
-esym(1726,endl,ends) // taking the address of an overloaded function
-esym(18,Data_t::Data_t) // definition not matching declaration
-elib(10) // expecting ')' -- comdef.h has a: #if defined( id
-elib(43) // vacuous array within _MIDL_FORMAT_STRING
-elib(602) // benign comment within comment
-elib(657) // declaring "anonymous struct" in union _LARGE_INTEGER
-elib(799) // long numerical constant for max. __int64
-elib(1502) // nothrow has no data members
-elib(1505) // no access specifier in base class specifier
-elib(1515) // AFX_THREAD_STATE member has no default constructor
-elib(1706) // Unusual declaration with a scope operator
-elib(1725) // data member is a reference
-elib(1548) // conflicting exception specifications
-elib(1737) // hiding global operator new
-elib(1739) // binary operator should be non-member function
-elib(1748) // non-virtual base class included twice
-elib(1759) // post-fix operator returns a reference
// Add elements of ole automation
lib-ole.lnt
// Options required for .net
-d_stdcall=__stdcall // make _stdcall equivalent to __stdcall
-d__interface=class // treat an interface the same as a class
-d__unaligned= // pass over the __unaligned keyword
-d__w64= // ignore this identifier
// Make sure you define your architecture type by unleashing one
// of the following defines:
-d_X86_
//-d_MPPC_
//-d_PPC_
//-d_IA64
//-d_MIPS
//-dMIDL_PASS
//-d_ALPHA_
-esym(40,DLGPROC) // used before being defined
-elib(146) // assuming binary constant
-elib(1015) // GetDefaultThreads not found in class
+ppw(using) // ignore #using for now.
-d__pragma(x)= // ignore the pragma extension
+rw(__thiscall,__ptr64) // additional qualifiers
-"d__identifier(x)=___identifier x"
// treat C++ keyword x as an identifier
-elibsym(1512) // base class destructor not virtual
-d_MSC_VER=1400 // compiler version (in decimal)
-d_MSC_FULL_VER{140050727} // full compiler version #
-d_MSC_EXTENSIONS{1} // Change value to 0 if not using extensions
-d_M_IX86=600 // assume Intel 80386 architecture -- modify to suit
// These two are defined only when the user compiles with the
// /Zc:wchar_t option for VC8. However, we have found
// -d_WCHAR_T_DEFINED created a rare problem. So, now, we just leave
// them undefined. If the user *really* wants to deal with the issues
// these options raise, they can uncomment them.
//-d_NATIVE_WCHAR_T_DEFINED{1}
//-d_WCHAR_T_DEFINED
-dinitonly= // Compiler should catch miss-uses.
// Lint can just skip over it.
-"dliteral=static const" // Documentation says they are
// equivalent for member data.
// When compiling with /J, use:
//+fcu // Plain char is unsigned
//-d_CHAR_UNSIGNED
// When compiling with /clr, /clr:pure or /clr:safe, use:
//-d__cplusplus_cli=200406 // for all 3
//-d_M_CEE_PURE // for /clr:pure
//-d_M_CEE_SAFE // for /clr:safe
//-d_MANAGED // for /clr
// When using any /clr form, use:
//-d_M_CEE
// When compiling with /GR, use:
//-d_CPPRTTI //Enable Run-Time Type Information
// When compiling with /GX, use:
//-d_CPPUNWIND //Enable Exception Handling
// When compiling with an /arch option, use:
//-d_M_IX86_FP=1 // For /arch:SSE
//-d_M_IX86_FP=2 // For /arch:SSE2
// When compiling for Win64, use:
//-d_WIN64
// When compiling with /Wp64, use:
//-d_Wp64 // 64-bit portability
// Be sure to define your platform if any of the following apply:
//-d_M_ALPHA // For DEC ALPHA platforms
//-d_M_IA64 // For Itanium 64-bit processors
//-d_M_X64 // For x64 processors
// When compiling with the /RTC option, use:
//-d__MSVC_RUNTIME_CHECKS // Using such checks
// When compiling with /openmp, use:
//-d_OPENMP=200203 // OpenMP specification date
// When compiling with /Zl, use:
//-d_VC_NODEFAULTLIB // Omit default library name in *.obj file
//__try_cast is like dynamic_cast (except the former throws where
//the latter returns 0).
-d__try_cast=dynamic_cast
//Partial support for the "old" (VC++ 2003) Managed Extensions
//syntax:
+rw( __gc, __value, __nogc, __pin )
-$ // $ can be used in identifiers
+rw( __allowed_on_parameter )
+rw( __allowed_on_function )
+rw( __allowed_on_typedecl )
+rw( __allowed_on_return )
+rw( __allowed_on_struct )
+rw( __allowed_on_function_or_typedecl )
+rw( __allowed_on_field )
+rw( __allowed_on_parameter_or_return )
+rw( __allowed_on_function )
//+rw( *type_traits ) //type traits support
//-header(co-msc80.h) // implicitly includes <typeinfo>. (Needed because
// MSVC 8 implicitly declares class type_info.)
env-vc8.lnt
/*
env-vc8.lnt: environment parameters for Microsoft's Visual C++ 8.0
If you are using Microsoft Visual Studio 2005 and you wish to invoke
PC-lint from that environment then add one or more commands to the
Tools menu as follows.
Simple Check
------------
For example, to add a simple facility to lint the current file you
may do the following:
1. From the Tools Menu choose "External Tools ..."
2. Click the "Add" button.
3. You will now be able to enter the fields of this Tool.
Modify them so that they approximate the following:
Title: PC-lint (Simple Check)
Command: c:/lint/lint-nt.exe
Arguments: -i"c:/lint" std.lnt env-vc8.lnt "$(ItemFileName)$(ItemExt)"
Initial Directory: $(ItemDir)
X_Use Output Window __Prompt for arguments __Close on exit
Please note that you will have to change the "Command:" path if the
PC-lint Installation Directory is anything other than c:/lint and
you will have to change the "Arguments:" line if the Configuration
Directory is anything other than c:/lint
4. Select OK to return to the main environment.
This will result in the Tools menu containing the additional item
"PC-lint (Simple Check)". Checking 'X' on 'Use Output Window' is
important because in this way you can advance from error to error
using the F8 key (Shift F8 to reverse).
Strings of the form $(...) are called macros and can be typed in
directly as shown or can be selected from a menu by clicking
a right arrow in the dialog box. $(ItemFileName) refers to the
file name of the currently edited file without its path and without
its extension. $(ItemExt) is its extension. $(ItemDir) represents
the file's directory.
You will probably want to advance your new tool upward into the initial
position of all tools while you are testing and modifying the command.
You can do this by using the "Move Up" button that appears on the External
Tools dialog.
The benefits of using "Initial Directory" are that file-names in lint
error messages will not be so long, and, also, this directory can contain
a std.lnt that overrides the global std.lnt in the Configuration Directory.
This Simple Check is fine to check stand-alone modules but to check
projects or to unit check modules that are in projects we need to
go a bit further ...
Project Creation
----------------
To lint an entire project we will need the names of all the modules
in the project. Visual Studio keeps these names (as well as some
appropriate options such as define options (-d...) and include options
(-i...) in a file named NAME.vcproj in the current project directory.
NAME is the name of the project and is identified by the macro
$(TargetName). PC-lint can read the .vcproj file and generate the
appropriate .lnt file. We recommend creating a tool for this purpose.
For this tool follow the steps 1-4 doing exactly the same thing as
above except in step 3, the information entered should be:
Title: PC-lint (Project Creation)
Command: c:/lint/lint-nt.exe
Arguments: -v -os("$(TargetName).lnt") "$(ProjectFileName)"
Init. Dir.: $(ProjectDir)
__Use Output Window __Prompt for arguments x_Close on exit
You will need to have an active project before this will work.
If you don't already have one you can obtain an active project
from the Solutions Explorer. You then click the newly added
"PC-lint (Project Creation)" tool on the tools menu to create
NAME.lnt.
The file created is an ASCII file and we recommend that you open it
within the IDE and examine it for any obvious flaws. This is your
chance to make any necessary modifications to the file as the process
of conversion may be less than perfect.
Project Check
-------------
Interestingly, by opening up the NAME.lnt file created above and
running the Simple Check described earlier you have the equivalent
of a full project check. However, we prefer to create a special
Project Check tool.
Now that we have a project file we can create a new tool called
"PC-lint (project check)". For this tool again follow steps 1-4 doing
exactly the same thing as above except in step 3, the information
entered should be:
Title: PC-lint (Project Check)
Command: c:/lint/lint-nt.exe
Arguments: -i"c:/lint" std.lnt env-vc8.lnt "$(TargetName).lnt"
Init. Dir.: $(ProjectDir)
X_Use Output Window __Prompt for arguments __Close on exit
Unit Check
----------
You can almost do a unit check on any single module by using the
Simple Check scheme suggested above. The only problems are that you
will need a -u option and you will not have the benefit of any -d or
-i options that have been placed into NAME.lnt created in the Project
Creation step. For this reason we suggest the following tool for
doing a unit check of any module that is part of a project and for
which a .lnt project file has been generated.
Title: PC-lint (Unit Check)
Command: c:/lint/lint-nt.exe
Arguments: -i"c:/lint" std.lnt env-vc8.lnt --u "$(TargetName).lnt" "$(ItemPath)"
Init. Dir.: $(ProjectDir)
X_Use Output Window __Prompt for arguments __Close on exit
Note that $(ItemPath) will provide a complete path name and in the
absence of a project.lnt file it would cause full path names to
appear in messages. But a side effect of using the project file
with the --u option means that we adopt the shorter names used
in the project file.
Suppressing Messages
----------- --------
Suppressing messages is normally done by adding message suppression
options to a file. For example, -e550 will suppress message 550.
There are numerous other options to suppress messages.
As the documentation indicates, the file
c:/lint/options.lnt
(where c:/lint/ is the Configuration Directory) is the presumed container
of your overall suppression policy. (Note: options.lnt is referenced
by std.lnt). Add a message suppression here and
you will affect all linting employing that configuration.
To suppress messages for a particular project (or for all projects
within a given project directory) you may do the following:
Create a file std.lnt that is contained in the project directory.
Make it refer back to the std.lnt in the Configuration Directory.
Then add additional message suppression options or indeed any options
you want. For example it might contain:
c:/lint/std.lnt // reference to original std.lnt
-e550 // project-specific option
In this way suppression is limited to a particular project.
Tool Bar
--------
You also have the option of creating a PC-lint toolbar within your Visual
C++ IDE. First, create one or more tools as described above. You
will need to know the number(s) of the tool(s) you want to place on
the tool bar. This can only be done by the painful and laborious
task of counting. Using the list provided by "Tools"/"External Tools",
jot down the numbers (starting with 1 at the top) of all the tools
to be added to the tool bar. We recommend placing all the PC-lint tools
on a single tool bar. Then select Customize from the Tools menu.
Select the Toolbars tab and click the New... button.
Give the Toolbar a name (E.g., PC-lint) in the dialog box provided and
click "OK". Confirm that the new toolbar is now floating on the
desktop and that a check has been placed in the check box next to the
new toolbar name. Then click on the Commands tab and in the Categories
box, scroll down to and click "Tools". In the Commands box, scroll down
to the appropriate external command number(s) corresponding to the
PC-lint commands you jotted down earlier. Drag the commands from the
Commands box to the toolbar. Their numeric name will change to the
logical name.
If you want to add a button image to the toolbar, you can choose one
via the Modify Selection button. Click Close and you now have your
own PC-lint for C/C++ button. (Note: If you change the location of
the PC-lint menu item on the Tools menu, you will change the subscript
and you will need to change the button(s) on the toolbar.)
*/
-"format=%(%F(%l):%) error %n: (%t -- %m)" // Messages will contain
// file information (%F), the line number (%l), the
// message number (%n), message type (%t) and message text (%m).
-hF2 // Make sure we ALWAYS provide file information ('F') and use 2
// lines (one for the source line in error and one for the message).
-width(0) // don't break messages at any particular width
-t4 // Presume that tabs are every 4 stops
+e900 // issue a message at termination.