Re: A plain approach - C / C++ / MFC Discussion Boards

Re: How to inverse a variable in bit level?

Randor 8-Jan-09 23:05

Randor

8-Jan-09 23:05

It works by masking bits/using the AND operator to get the bits in the proper location. It is very beautiful, you can see it working here:

DWORD iTmp =0;
BYTE b = 139;			// 00000000 00000000 00000000 10001011

DWORD c = (b * 0x0802LU);	// 00000000 00000100 01011001 00010110
c &= 0x22110LU;		// 00000000 00000000 00000001 00010000

DWORD d = (b * 0x8020LU);	// 00000000 01000101 10010001 01100000
d &= 0x88440LU;		// 00000000 00000000 10000000 01000000

iTmp = (c | d);		// 00000000 00000000 10000001 01010000
iTmp = iTmp * 0x10101LU;	// 10000001 11010001 11010001 01010000

iTmp >>= 16;			// 00000000 00000000 10000001 11010001
b = iTmp;			// 00000000 00000000 00000000 11010001

Best Wishes,
-David Delaune

Re: How to inverse a variable in bit level?

toxcct8-Jan-09 23:49

toxcct

8-Jan-09 23:49

definitely marvelous... with your accordance, may I quote you in my article which talks about bits twiddling ?

[VisualCalc][Binary Guide][CommDialogs] | [Forums Guidelines]

Re: How to inverse a variable in bit level?

Randor 8-Jan-09 23:55

Randor

8-Jan-09 23:55

toxcct wrote:
may I quote you

Hi toxcct,

Absolutely. Smile | :)

Best Wishes,
-David Delaune

Re: How to inverse a variable in bit level?

toxcct9-Jan-09 0:00

toxcct

9-Jan-09 0:00

shame on me, it was already (partially) there
check the article[^], at the Links chapter (down the end) ^^

there it is in the "Reversing bit sequences" chapter of this page[^]...

[VisualCalc][Binary Guide][CommDialogs] | [Forums Guidelines]

Re: How to inverse a variable in bit level?

Randor 9-Jan-09 0:09

Randor

9-Jan-09 0:09

Hi,
That is an excellent link to an awesome collection of bit manipulation codes. I have used many of them in various projects over the years. Smile | :)

Best Wishes,
-David Delaune

Re: How to inverse a variable in bit level?

toxcct10-Jan-09 22:57

toxcct

10-Jan-09 22:57

done... the article[^] has been updated Smile | :)

[VisualCalc][Binary Guide][CommDialogs] | [Forums Guidelines]

Re: A plain approach

CPallini8-Jan-09 22:44

CPallini

8-Jan-09 22:44

Inefficient, sure, but maybe easier to grasp.

BYTE b = 139;
BYTE r = 0;

for (int i=0; i<8; i++)
{
  r = r << 1;
  r |= b & 0x1;
  b = b >> 1;
}

If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.

This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke

[My articles]

Re: A plain approach

Iain Clarke, Warrior Programmer9-Jan-09 0:16

Iain Clarke, Warrior Programmer

9-Jan-09 0:16

I'm not sure it's much less efficient - all those multiplies will quite a high cost.

If it's just a BYTE, you could cut it down more by not even bothering with a loop:

BYTE b = 139;
BYTE r;
r = ((b & 1) << 7) | ((b & 2) << 5) | ((b & 4) << 3)  | ((b & 8) << 1)  | ((b & 16) >> 1)  | ((b & 32) >> 3)  | ((b & 64) << 5)  | ((b & 128) >> 7) )

This would not extend well for unsigned __int64...

Iain.

Codeproject MVP for C++, I can't believe it's for my lounge posts...

Re: A plain approach

CPallini9-Jan-09 0:28

CPallini

9-Jan-09 0:28

Uhm...loop unrolling: the oldie-goldie game-hacker Iain!
Big Grin | :-D

Re: A plain approach

Randor 9-Jan-09 1:05

Randor

9-Jan-09 1:05

Your version should be slightly faster. In theory on the Pentium 4 instruction cycle count should be as follows:

Bit masking technique:
3 MUL * 16 cycles
1 OR * 0.5 cycle
2 AND * 0.5 cycle
1 SHR * 4 cycles
===============
53.5 cycles

 
Your unrolled shifting:
8 AND * 0.5 cycles
7 OR * 0.5 cycles
5 SHL * 4 cycles
3 SHR * 4 cycles
================
39.5

It would be interesting to have MSVC output an .asm file with the /FA option. It would be interesting to see what the 2005/2008 compiler optimizer does to all of our codes!

Best Wishes,
-David Delaune

Re: A plain approach

CPallini9-Jan-09 1:43

CPallini

9-Jan-09 1:43

I think you may drop off a pair of ANDs in Iain expression.

BTW: very good job, I like soo much your posts.
Smile | :)

Re: How to inverse a variable in bit level?

Luc Pattyn9-Jan-09 2:51

Luc Pattyn

9-Jan-09 2:51

Hi,

if you need it once, the answers given to you are fine.
if you need to bit-swap an entire file or large memory buffer, I suggest you create a 256-byte lookup table, which you initialize by running one of the given code snippets for all possible byte values.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

Love, happiness and fewer bugs for 2009!

Potential problem with 'typedef redefinition'

6Qing888-Jan-09 20:27

6Qing88

8-Jan-09 20:27

hi, everyone
recently I started a new project which needs to be written in plain C,
and I got problem with different compilers that:

I want the declaration of public APIs along with data types required to
be in some separate files, and the definition of the types in some others,
for example:

interface.h
struct _VECTOR;
/**
* there's a problem with the absence of 'struct' in function
* declaration at least in mingw2.95, but that's not the point
* of my post.
*/
void _InitVector(struct _VECTOR* pVector);

implementation.h
#include "interface.h"

/* my point is here */
typedef struct _tagVECTOR {
int a;
} _VECTOR, *_PVECTOR;

implementation.c
#include "implementation.h"

void _InitVector(struct _VECTOR* pVector)
{
pVector->a = 1; /* derefferencing pointer to imcomplete type. */
}

It seems that the compiler cannot recognize the fact that _VECTOR is a typedef
name of the struct _tagVECTOR;

By the definition of 'typedef', I know that _VECTOR declared in "interface.h"
is exactly the same entity to that in "implementation.h", although it turned
out to be a typedef name of _tagVECTOR, but it won't compile in mingw2.95 with
the error mentioned in the code above.

Revised version can solve the problem:
typedef struct _VECTOR {
int a;
} *_PVECTOR;

But I want it to be consistent with the general idea, so I revised further:
typedef struct _VECTOR {
int a;
} _VECTOR, *_PVECTOR;

It compiled with no problem, but I'm not sure if I'm lucky enough with other
compilers holding the potential issue of "typedef redefinition".

Thanks in advance.

Smile | :)

Re: Potential problem with 'typedef redefinition'

sashoalm8-Jan-09 20:32

sashoalm

8-Jan-09 20:32

Why are you posting a plain C question in a C++ / MFC forum?

That said, I think you should write
void _InitVector(_VECTOR* pVector)
instead of
void _InitVector(struct _VECTOR* pVector)

There is sufficient light for those who desire to see, and there is sufficient darkness for those of a contrary disposition.
Blaise Pascal

Re: Potential problem with 'typedef redefinition'

6Qing888-Jan-09 20:40

6Qing88

8-Jan-09 20:40

hi,
The joke is that because I think C is a subset of C++. Smile | :)

And I got problem with the absence of that 'struct':
[Error] parse error before '*'
It seems the pre-declaration can never be sighted by the compiler(mingw2.95). Frown | :(

Re: Potential problem with 'typedef redefinition'

sashoalm8-Jan-09 21:37

sashoalm

8-Jan-09 21:37

6Qing88 wrote:
And I got problem with the absence of that 'struct

Sorry I didn't read your post carefully. The problem is that you declare _VECTOR as a struct first

struct _VECTOR;
void _InitVector(struct _VECTOR* pVector);

but as a typedef later

typedef struct _tagVECTOR {
int a;
} _VECTOR, *_PVECTOR;

You can't forward declare a typedef as a struct, because it is not a struct - it is a typedef. I think it is intended to be that way (i.e. it's not the compiler but the C language specs that require it).

You can try writing this at the header (I haven't tested it)

struct _tagVECTOR;
typedef struct _tagVECTOR _VECTOR;
void _InitVector(_VECTOR * pVector);

There is sufficient light for those who desire to see, and there is sufficient darkness for those of a contrary disposition.
Blaise Pascal

Re: Potential problem with 'typedef redefinition'

Stuart Dootson8-Jan-09 21:45

Stuart Dootson

8-Jan-09 21:45

Your problem is that you're using struct _VECTOR in interface.h and _VECTOR == struct _tagVECTOR elsewhere. Those are two different types. It doesn't matter that you use _VECTOR in both - struct tag names are in a different namespace than type names.

Thus, in implementation.c, struct _VECTOR is incomplete. struct _tagVECTOR, aka _VECTOR is complete...but you are not referencing that type!

Font and size of default message Box

tasumisra8-Jan-09 20:04

tasumisra

8-Jan-09 20:04

Hi all,
I am making look and feel of a Dialog same as message box can anybody tell me what font and size should i make it so that it should look exactly similar as messageBox.

Thanks

vikas da

Re: Font and size of default message Box

Randor 8-Jan-09 20:34

Randor

8-Jan-09 20:34

You should use the MS Shell Dlg or MS Shell Dlg [^] logical fonts to have the same behavior of MS Windows dialogs. I am pretty sure that the default font size is 8 for most english users but if I remember correctly the default font size can be larger in an international installation.

Also keep in mind that although most users with 20-20 vision have the DPI set at 96, there may be individuals with DPI set at 120. You should test your messagebox dialog to ensure that it can handle all environments.

Best Wishes,
-David Delaune

Load data in Combobox in VC++6.0

aa_zz8-Jan-09 19:12

aa_zz

8-Jan-09 19:12

guide me load data to combobox for vc++6.0. please. thanks

"
enum { IDD = IDD_SPOOLIMAGE_DS_PP };//gọi form
DDX_Control(pDX, IDC_TEST, value_del);

CComboBox* cb = (CComboBox*)GetDlgItem(IDC_TEST);
cb->Clear();

CString sName = "";
for(int i=0; i<10;i++)
{
cb->SetItemData(i, DWORD(i));
}

"

Re: Load data in Combobox in VC++6.0

Hamid_RT8-Jan-09 19:20

Hamid_RT

8-Jan-09 19:20

And what the problem?

Of one Essence is the human race
thus has Creation put the base
One Limb impacted is sufficient
For all Others to feel the Mace
(Saadi )

Re: Load data in Combobox in VC++6.0

aa_zz8-Jan-09 19:24

aa_zz

8-Jan-09 19:24

can you tell specification ? thanks . wish your help

Re: Load data in Combobox in VC++6.0