|
Mark Salsbery wrote: GetCommandLine()
Excellent! That along with Mathew's suggestion to use a windows application is just what I need to get rid of the flicker.
|
|
|
|
|
Can we overload a function by return type?
|
|
|
|
|
|
The simple answer is no. Two functions which differ only by return type are not allowed.
You can make a function return a template defined type but unless that type is also used for one of the parameters you're going to find it hard to get the compiler to recognise which type you intend to replace your template argument with in any given case. some experimentation required I think.
You could consider passing in the return value by reference and then you'd have no problem doing ordinary overloading.
int Myfunc(void)
{
}
float Myfunc(void)
{
}
void Myfunc(int& iRef)
{
}
void Myfunc(float& fRef)
{
}
Nothing is exactly what it seems but everything with seems can be unpicked.
|
|
|
|
|
No. Because when you call the function the compiler won't be able to know which overload you're calling.
|
|
|
|
|
In general no. The only exception (I believe) is casting operators.
Steve
|
|
|
|
|
english word for base-2 number is binary, for base-8 and base-16 are octal and hexadecimal respectively.
but:
which english word is for base-4 number?
i can not find it in google, maybe you know it.
thx.
|
|
|
|
|
|
The more interesting question is:
If "bit" is the abbreviation for "binary digit", what would be the abbreviation for "ternary digit"? 
|
|
|
|
|
trits 
This episode brought to you by the letter N
|
|
|
|
|
Then we will have to call "bits" "brits".
|
|
|
|
|
in Chinese they're simply called "base-2 number", "base-3 number", "base-4 number", etc. So maybe Chinese isn't as difficult as it seems~
|
|
|
|
|
I cant get SACL of an existing file on my system with this code.
GetNamedSecurityInfo returns 0 (ERROR_SUCCESS) , which means
there is no error in execution of it but pSACL still becomes 0x00000000
and IsValidAcl(pSACL) returns FALSE .
#include <windows.h>
#include <aclapi.h>
unsigned long enable_privilege(const char *priv)
{
HANDLE token = INVALID_HANDLE_VALUE;
u status = OpenProcessToken(
GetCurrentProcess(),
TOKEN_ADJUST_PRIVILEGES,
&token
);
if( !status )
{
return GetLastError();
}
unsigned char buf[sizeof(TOKEN_PRIVILEGES) + sizeof(LUID_AND_ATTRIBUTES)];
TOKEN_PRIVILEGES *privs = (TOKEN_PRIVILEGES*)buf;
status = LookupPrivilegeValue(
NULL,
priv,
&privs->Privileges->Luid
);
if( !status )
{
u err = GetLastError();
CloseHandle(token);
return err;
}
privs->PrivilegeCount = 1;
privs->Privileges->Attributes = SE_PRIVILEGE_ENABLED;
AdjustTokenPrivileges(
token,
FALSE, // do not disable all
privs,
0, // zero buffer for prev state
NULL, // prev state don't care
NULL // no sink for returned prev state size
);
status = GetLastError();
CloseHandle(token);
return status;
}
int _tmain(int argc, _TCHAR* argv[])
{
PACL pSACL = NULL;
PSECURITY_DESCRIPTOR pSD = NULL;
unsigned long returnValue ;
enable_privilege(SE_SECURITY_NAME );
returnValue = GetNamedSecurityInfo(
"D:\\aa.txt", // object name
SE_FILE_OBJECT, // object type
SACL_SECURITY_INFORMATION, // information type
NULL, // owner SID
NULL, // primary group SID
NULL, // DACL
&pSACL, // SACL
&pSD); // SD
if(!IsValidAcl(pSACL))
printf("FAILED , ERROR CODE : %d\n\n",returnValue);
LocalFree(pSD);
return 0 ;
}
|
|
|
|
|
Hi, I'm new to Visual C++ 2005, and I'm trying to create a simple "hello world" kernel. Can I use Visual C++ 2005? I've tried "Win32 Console Application", the PE result still call Windows dependent library. Are there some arguments I must pass to "CL.EXE" so I can get clean "PE EXE"?
Can "LINK.EXE" create another binary interface except PE/COFF? And why we cannot relocate the PE base location to a value below 64K? I still new to PE too.
Btw, I heard about "ML.EXE". Is that full version of MASM packaged together with Visual C++?
|
|
|
|
|
If by 'kernel' you mean something that runs without another operating system present, you can't do it in Visual C++ 2005 unless you already have a bootloader that gets the system into 32-bit (or 64-bit) protected mode. The PC traditionally starts up in 16-bit Real Mode - you'll need a 16-bit real mode compiler to write code for this mode. Microsoft's last 16-bit compiler was Visual C++ 1.52c, which is available to MSDN subscribers. I believe Extensible Firmware Interface switches to 64-bit protected mode within the early ROM boot code - 32-bit EFI exists but Microsoft aren't supporting it for Windows Vista (64-bit EFI on x64 is due to be supported in Windows Vista SP1 and Windows Server 2008).
To stop the compiler and linker producing an executable that links to any Win32 DLLs, you need to tell it not to link the Microsoft C Runtime. You do this with the /nodefaultlib switch. You will also need to change the entry point function; by default this is (for an ANSI console application) mainCRTStartup. You need to supply your own with the /entry switch. If you were producing a Windows binary the entry point function would need to match the WinMain function.
LINK only supports generating 32-bit (or 64-bit) PE/COFF outputs. It can accept Borland's Object Model Format files but not create them. Windows only runs PE files - even the Windows kernel, NTOSKRNL.EXE, is a PE file, as are device drivers (they're linked /DRIVER /SUBSYSTEM:NATIVE).
The PE base address cannot be below 64KB as the first 64KB of the address space is reserved in Windows. This is so that functions that take a resource identifier can differentiate between a numeric identifier (a value below 65536 cast to an LPTSTR) and a string identifier.
ML is the macro assembler - it stopped being a separate product a while ago. Again, it's designed to produce PE/COFF files. I'm not sure which Visual Studio editions it ships with - it might not be in Express Edition.
|
|
|
|
|
Mike Dimmick wrote: you can't do it in Visual C++ 2005 unless you already have a bootloader that gets the system into 32-bit (or 64-bit) protected mode.
I just realized why some CALL in code section use 4 bytes displacement instead of 2 bytes. Thanks, I'll create a code to switch to protected mode first before calling the PE. Because the PE is memory-mapped, I think I can just load the PE into memory and jump into it entry-point from image-base.
Thanks.
|
|
|
|
|
Ok a few bits here where I might be able to contribute:
SolidusSnake wrote: the PE result still call Windows dependent library.
If you're still linking to the C Runtime which VS 2005 does pretty silently then you'll get loads of dependencies. If you remove the CRT link and set /nodefaultlib you can get it down to a few dependencies on Kernell.dll/NtDll.dll.
SolidusSnake wrote: Can "LINK.EXE" create another binary interface except PE/COFF?
There are alternatives to COFF on the symbol side the same as there were in VC6 but you're stuck with PE as far as I know.
SolidusSnake wrote: I heard about "ML.EXE". Is that full version of MASM packaged together with Visual C++?
It certainly is a working version of MASM, whether it's 'full' or not I don't know as it's the first MS assembler I've used.
SolidusSnake, where have I come accross that name before?
Nothing is exactly what it seems but everything with seems can be unpicked.
|
|
|
|
|
Thanks for the answer.
Matthew Faithfull wrote: SolidusSnake, where have I come accross that name before?
That from my favorite game, Metal Gear Solid [Hideo Kojima]. There are Solid and Solidus Snake.
|
|
|
|
|
heh,
Your post caused me to start experimenting, and heres the best I could come up with:
2 kilobytes with VC8 
Make sure [Code Generation] -> [Buffer Security Check] is set to NO. (I could not find a pragma to accomplish this)
<br />
#include windows.h<br />
<br />
#define WIN32_LEAN_AND_MEAN<br />
<br />
#pragma comment(linker, "/NODEFAULTLIB")<br />
#pragma comment(linker, "/SUBSYSTEM:console")<br />
#pragma comment(linker, "/MERGE:.rdata=.data")<br />
#pragma comment(linker, "/MERGE:.text=.data")<br />
#pragma comment(linker, "/STACK:65536,65536")<br />
#pragma comment(linker, "/OPT:NOWIN98") <br />
<br />
#pragma comment (linker, "/ENTRY:main")<br />
int __cdecl main()<br />
{<br />
char msg[] = "Hello World";<br />
HANDLE stdout = GetStdHandle(STD_OUTPUT_HANDLE);<br />
DWORD dwCount;<br />
WriteFile(stdout, msg, lstrlen(msg), &dwCount, NULL);<br />
return 0;<br />
}<br />
The file would be slightly smaller if we changed the default alignment.
Best Wishes,
-Randor (David Delaune)
|
|
|
|
|
Thanks for the experiment and the result here.
|
|
|
|
|
I'm talking about win32 API here.
My example:
There was 2 available languages in the system( English and French for
example), app1 was launched (notepad for example), then I launched new
process - app2 which calls LoadKeyboardLayout() ( loads German for example)
and displays some messagebox, after that I launched app3 (another instance
of notepad). So if I set focus on app2 or app3 - language bar displays 3
languages (English, French, German), but if I focus app1 - language bar
displays only two languages (English, French). So I conclude that language
list of processes created before LoadKeyboardLayout() call was not updated.
If I for example use "Text Services and Input Languages" GUI instead of
LoadKeyboardLayout() call to add German, language bar displays 3 available
languages either I focus app1, app2 or app3.
So, when I use "Text Services and Input Languages" GUI to add language,
system updates all running processes, while LoadKeyboardLayout() updates
only processes started after the call. So system does "more" then
LoadKeyboardLayout().
What ?
|
|
|
|
|
Generally when you change a system setting you should broadcast the WM_SETTINGCHANGE message to inform other processes about a change of settings, and allow them to reload any modified settings.
|
|
|
|
|
not bad idea,
but do not work.
|
|
|
|
|
I have a view class derived from CFormView class.I have used scroll bar by using 'SetScrollSizes(...)' function. Now the problem is that when I set the Scroll Bar position by using 'SetScrollPos(...)' function, the view postion is not changed according to the Scrollbar positon. But if I change the Scrollbar position by using mouse, the view posion is changed accroding to the Scrollbar position. I have used also used OnVerScroll(...) and OnHorScroll(...) message mapping functions to solve the problem but it does not work.
|
|
|
|
|
I don't think SetScrollPos() causes a WM_HSCROLL/WM_VSCROLL message to be sent, so after you
call SetScrollPos() you should redraw the scrolled window contents accordingly.
Mark
This episode brought to you by the letter N
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
