C / C++ / MFC

John R. Shaw19-Mar-06 12:22

19-Mar-06 12:22

I would need to look it up, because I never understood why it would be classified as undefined behaviour.

D'Oh! | :doh:

My answer was very weak and I should not have attempted it [without research], becuase it actually applies to the "old argument" below, following my explination of what I ment.

I said "at the least" because if it is undefined behaviour then it would be compiler dependent (non-portable) and a warning should be generated to inform the programmer of this.

I said "at the most" because if it is undefined behaviour then it would be compiler dependent and an error would force the programmer not to write non-poratable code.

I believed it was consider undefined behaviour in the passed, which means the compiler vender could decide what behaviour to apply, and still be fully compliant. This may no longer be the case, as the behaviour may now be deifined.

Here is the old argument:

int i = 0;<br />
i = i++;

Should i now be equal to 0 or 1?

My answer and the answer Microsoft chose is 1, which I think is the only logical answer. But there where those who disagreed and said the answer should be 0, because they believed that was the intent of the progammer.

So if you throw out logic and just consider the intent, then the language would not be able to define the behaviour. I think the real problem was that different venders had already implimented different solutions, before the standared was complete.

When I originaly gave my answer of 1 to that question (years ago), I was informed that my answer was wrong, becuase it was undefined and therefore could be either 0 or 1.

VS6 may be fully compliant for C (?), but it is not fully compliant with the C++ standard.

VS2003, according to Microsoft, is a full ASCII/ISO C++ Standards compliant compiler.

INTP
Every thing is relative...

Ryan Binns12-Mar-06 17:31

Ryan Binns

12-Mar-06 17:31

John R. Shaw wrote:
I have not looked at the sights you provided, but that is not an error. A warning might not be a bad idea for that, but come on (it's legal code).

Actually, it does violate the standard:

"Between the previous and next sequence point an object shall have its stored value modified at most once by the evaluation of an expression. Furthermore, the prior value shall be accessed only to determine the value to be stored."

The second sentence is the killer here. I think it's a silly statement, but perhaps that's just me. To me, it would be more intuitive to state that an object's value is not changed until the compiler reaches a sequence point - nice and clean and predictable, without preventing totally unambiguous code such as this example.

Ryan

"Punctuality is only a virtue for those who aren't smart enough to think of good excuses for being late" John Nichol "Point Of Impact"

John R. Shaw16-Mar-06 15:44

16-Mar-06 15:44

Ryan Binns wrote:
Actually, it does violate the standard

Actually, undefined behaviour is not a violation of the standard. It is what it says "undefined", that means it is up to the compiler [vender] to determine what its behaviour is.

Personaly, I think it should translate into:

data[i] = i;
++i;

data[i] = (temp = i + 1);
i = temp;

but that is just my opinion.

INTP
Every thing is relative...

Trollslayer12-Mar-06 1:45

Trollslayer

12-Mar-06 1:45

Did you change the warning level to 4? (Project->Settings->C/C++)

Elaine Rose | [Rose]

The tigress is here Big Grin | :-D

Chintoo72312-Mar-06 3:44

Chintoo723

12-Mar-06 3:44

Trollslayer wrote:
Did you change the warning level to 4? (Project->Settings->C/C++)

I just checked with both VS6 and VS 2003, both of them dont detect and warn for a[i] = i++. Do you have VS 2005 to check this out? GCC 3.4.3 does, if you want to check this out with GCC.

thanks!

How to translate applications using resources files

Trollslayer12-Mar-06 10:06

Trollslayer

12-Mar-06 10:06

I just tried on VS2005 with warning level 4 and it wasn't reported.

The tigress is here Big Grin | :-D

Allad11-Mar-06 17:40

Allad

11-Mar-06 17:40

Hello,
I'd like to release an application in several languages(english, french and spanish).
I know I can do it by traducing only my resource file. In fact, I've included all strings I use as resources.
But I don't want to duplicate my application directory for each language. I'd like to manage all languages only from one workspace under Visual C++ 6.
Can you please tell me how to do or point me to a document discussing it?

Thanks

Best regards,
Allad

----
Navigator - Your best alternative to Windows Explorer

Re: How to translate applications using resources files

John R. Shaw11-Mar-06 22:32

Re: How to translate applications using resources files

11-Mar-06 22:32

I have written programs that allow multiple languages, but I have not written one that supported multiple languages for the interface.

Given that, there is a book from Microsoft press that explains the subject. I forget its name but it was included in the VC6 MSDN library and the 2nd edition came out last year.

There is also an article some where on CP on the subject.

Essentially it breaks down to providing a DLL for each language that contains the strings for the required languages. In order to change the language all you have to do is change the DLL. In other words, your string resources are stored in a DLL. You still have to make sure that you give enough room in the text fields (at least in dialogs) for longer text strings or shorter text strings.

The biggest problem is working with Asian text, because you have to give extra room for the height. Outside of Asian characters I have not [yet] ran across double byte characters.

I do need to warn you, in the middle-east we may have a totally different problem, but I am not going to discuss that now.

INTP
Every thing is relative...

Allad12-Mar-06 2:00

Allad

12-Mar-06 2:00

Thank you for replying.
However, I do not want to write an application that supports multiple language for the interface.
I'd just like to build one different version of my app for each language. Maybe it's a bit easier?

Best regards,
Allad

----
Navigator - Your best alternative to Windows Explorer

Re: How to translate applications using resources files

John R. Shaw16-Mar-06 15:24

How can i enumerate all users in PC?

16-Mar-06 15:24

Sorry, I have been on a break.

The simplist approach would be to write your code using the functions prefixed with "_t" (Microsoft specific) to support single-byte, multibyte, and UNICODE character sets, depending on how it is compiled. Then require that their machine requires an OS (NT, Win2000, WinXP+) that supports UNICODE and only compile your code for UNICODE. Then you should be able to create one application that supports all UNICODE supported languages with one program (not including the interface).

If you need to support Win3.1, Win95, Win98, or Millinium Edition then it is to complicated to explain.

Good luck!

INTP
Every thing is relative...

fcoder11-Mar-06 16:41

fcoder

11-Mar-06 16:41

I need to delete application files producing when user use my App in the uninstalling process.How can i know all users' account in order to delete files in Document Setting(Win2k).

Thanks

Anton Mikhalyov11-Mar-06 21:13

11-Mar-06 21:13

NetUserEnum
The NetUserEnum function provides information about all user accounts on a server.

Hamid_RT11-Mar-06 21:14

Hamid_RT

11-Mar-06 21:14

Hi fcoder,
I think http://msdn.microsoft.com/library/default.asp?url=/library/en-us/ad/ad/enumerating_users.asp[^]
is helpful for you

Hamid_RT11-Mar-06 21:24

Hamid_RT

11-Mar-06 21:24

PUSER_INFO_3 users;
LPWSTR lp1;
DWORD d1,d2;

DWORD SizeName=sizeof(Name);
GetComputerName(Name,&SizeName);
lp1=A2W (Name);

NetApiBufferAllocate(1024,(LPVOID*)&users);

NetUserEnum(lp1,3,FILTER_NORMAL_ACCOUNT,(LPBYTE*)&users,MAX_PREFERRED_LENGTH, &d1,&d1,NULL );

and also use USES_CONVERSION; in the first code

I hope helpful for you

John R. Shaw11-Mar-06 22:42

11-Mar-06 22:42

You need to reword this. I am amazed I understood what I did.

INTP
Every thing is relative...

method of hook dll

HOW WHAT11-Mar-06 15:15

11-Mar-06 15:15

i want to hook the dll funtion, my mothod is modification the exe Import Table.

have other method of hook dll and not need modification the Import Table and Export Table.

thanks.

Anton Mikhalyov11-Mar-06 18:17

11-Mar-06 18:17

HOW WHAT wrote:
have other method of hook dll and not need modification the Import Table and Export Table.

you can change first bytes of function to far jmp that points on your code.

HOW WHAT11-Mar-06 18:41

11-Mar-06 18:41

how to ? ex.?

Anton Mikhalyov11-Mar-06 20:52

11-Mar-06 20:52

This example shows the hook of GetModuleHandleW in target process(process where you injected your dll). It works only in 32-bit windows.
If you injected your dll into multithreaded process you must synchronize execution of this code with other threads or program sometimes may crash.

<br />
#define BYTES_COUNT 0x05<br />
<br />
BYTE firstBytes[BYTES_COUNT];<br />
void *pfnGetModuleHandle;<br />
HANDLE __stdcall Handler(HANDLE hModule);<br />
<br />
// Only running process hook<br />
BOOL HookGetModuleHandle()<br />
{<br />
	HANDLE hKernel32;<br />
	DWORD dwOldProtect;<br />
<br />
	hKernel32 = LoadLibrary(_T("kernel32.dll"));<br />
	pfnGetModuleHandle = GetProcAddress((HMODULE)hKernel32, "GetModuleHandleW");<br />
<br />
	if (hKernel32 == NULL || pfnGetModuleHandle == NULL)<br />
	{<br />
		return FALSE;<br />
	}<br />
<br />
	memcpy(&firstBytes, pfnGetModuleHandle, BYTES_COUNT);<br />
<br />
	if (!VirtualProtect(pfnGetModuleHandle, BYTES_COUNT, PAGE_READWRITE, &dwOldProtect))<br />
	{<br />
		return FALSE;<br />
	}<br />
<br />
	*((BYTE*)pfnGetModuleHandle) = 0xE9;<br />
	*((DWORD*)(((BYTE*)pfnGetModuleHandle)+1)) = (DWORD)Handler - (DWORD)pfnGetModuleHandle - BYTES_COUNT;<br />
<br />
	if (!VirtualProtect(pfnGetModuleHandle, BYTES_COUNT, dwOldProtect, NULL))<br />
	{<br />
		return FALSE;<br />
	}<br />
<br />
	return TRUE;<br />
}<br />
<br />
BOOL UnhookGetModuleHandle()<br />
{<br />
	DWORD dwOldProtect;<br />
<br />
	if (!VirtualProtect(pfnGetModuleHandle, BYTES_COUNT, PAGE_READWRITE, &dwOldProtect))<br />
	{<br />
		return FALSE;<br />
	}<br />
<br />
	memcpy(pfnGetModuleHandle, &firstBytes, BYTES_COUNT);<br />
<br />
	if (!VirtualProtect(pfnGetModuleHandle, BYTES_COUNT, dwOldProtect, NULL))<br />
	{<br />
		return FALSE;<br />
	}<br />
<br />
	return TRUE;<br />
}<br />
<br />
HANDLE __stdcall Handler(HANDLE hModule)<br />
{<br />
	HANDLE returned;<br />
	printf("GetModuleHandleW call detected\n");<br />
	UnhookGetModuleHandle();<br />
	__asm {<br />
		push hModule<br />
		call pfnGetModuleHandle<br />
		mov returned, eax<br />
	}<br />
	HookGetModuleHandle();<br />
}<br />

HOW WHAT12-Mar-06 1:24

12-Mar-06 1:24

your method and modification the Import Table, which will be highness and safe?

Anton Mikhalyov12-Mar-06 3:55

12-Mar-06 3:55

My method is better to use, because it never fail, except when application detected and removed hook(very very rarely), but it creates additional problems with synchronization in mt applications and it doesn't work on windows other than 32 bit(without modification).
IAT modification is also good method, but this hook wouldn't work if application directly calls functions(call GetProcAddress to retrieve address of function...). Second method fails more offen than first.
Also you can create a dll that debugs process in which it injected by inserting int3 instructions into functions you want to hook. If number of functions to hook =< 4 you can use hardware breakpoints, setting hardware breakpoints do not require any code modifications.

HOW WHAT12-Mar-06 4:52

12-Mar-06 4:52

1.Your method if had some call hooked function may be happen some exception, right?

2.How to do like you say
"Also you can create a dll that debugs process in which it injected by inserting int3 instructions into functions you want to hook. If number of functions to hook =< 4 you can use hardware breakpoints, setting hardware breakpoints do not require any code modifications. "

Some time ago, I download codeproject about use debug process to inject dll, but in win98 all application will happen crash.

Anton Mikhalyov13-Mar-06 1:17