|
Member 7989122 wrote: read/write the "array element" at 0[address]
The compiler will be quite happy, but the hardware will raise an exception at run time (unless you are on an old 8080 level system).
BTW, some excellent explanations in your replies. 
|
|
|
|
|
Additional comment: In the 1980s, there was an intense discussion about whether a data item should provide information about its type or not. If the data item knows whether it is 8, 16, 32 or 64 bits, float or integer, or whatever, it is represented once, not in every single instruction operating on it. The instruction code would not need different formats (/additional bits) to distinguish between e.g. integer add and floating add. So instructions - of which there are usually a lot more than values - would be a lot more compact, and the risk of interpreting a value in the wrong format significantly reduced.
An essential element of this approach was to store index limits once, with the array. If every array access makes is own check against array limits, the code/data for this check could grow significantly - way beyond what you are aware of when reading the code for a single access. It looks so tempting, 'I do it only when there is a risk of exceeding the array limits, but save the check when there is not'. I am 100% sure that the compiler (or even runtine system) is a lot more clever than you to make that decision. In languages with runtime array (/object) descriptors, it is possible. In C, or C++ in classical C usage pattern, it is quite difficult.
|
|
|
|
|
Member 7989122 wrote: In languages with runtime array (/object) descriptors, it is possible. In C, or C++ in classical C usage pattern, it is quite difficult.
The proprietary language in which I worked for many years (designed in the late 1970s) had descriptors built in. The compiler knew the size of each array element, and a descriptor was a pointer to the beginning of the array (as in C++) and its size (number of elements). The compiler then generated code to perform bounds checking on an index at run-time.
If the size of the array was known at compile time, a descriptor wasn't needed; the compiler just did bounds checking against the fixed length. Descriptors were used for dynamically allocated arrays or to reference a subset ("slice") of a larger array.
Much more recently, C++ has added std::span[^] to the STL, which does the same thing.
modified 12-Apr-20 9:46am.
|
|
|
|
|
It is a historic fact, now days on C99, C11 and beyond you might more correctly use a uintptr_t.
All you want really is the address but how big is that address it could be 16bits on a small micro controller, 32 bit on large CPU or 64bits on 64bit cpu. A void* pointer was usually big enough to ensure it had enough bits to point to any valid address on the CPU. So the size of a void* is completely compiler dependent. Back in the day there were also other features a void* could be cast to and from any other pointer without warning. The reason is obvious you want to be able to copy the address to a pointer of type and use normal c pointer functions.
Now move forward and look at a uintptr_t this is the C99 definition in stdint.h for portability under XSI-conformant systems
Quote: "an unsigned integer type with the property that any valid pointer to void can be converted to this type, then converted back to pointer to void, and the result will compare equal to the original pointer".
It acts like a void* pointer with one safety added the conversion to and back is guaranteed, which never was with void* and occasionally cropped up and you can safely do it on any pointer type.
It also means when you look at it in a debugger it shows as an unsigned integer rather than a pointer which is more in keeping with what it really is ... an address to somewhere in CPU memory space.
In vino veritas
|
|
|
|
|
Member 7989122 thanks for sharing that. Quote: the pointer itself is copied to another pointer
this stuck in my mind
|
|
|
|
|
C++11 or previous.
Not C#.
Not Visual C++.
Not .NET anything.
Hello.
I have the following code. It shows a blank listbox. I would like to be able to use multiple listboxes, etc. and put them in a buffer to show later.
Handle_of_ListBox = CreateWindowEx(
WS_EX_CLIENTEDGE, L"LISTBOX", nullptr,
WS_CHILD | WS_VISIBLE, MainWindow_Width - 400, MainWindow_Height - 300, 300, 200, Handle_of_MainWindow,
nullptr,
hInstance,
nullptr
);
I can use the following to show or hide the listbox in the winmain, but I want to show or hide it in a pre-screen buffer.
I use the following to initialize a buffer.
void Initialize_FRONT_BUFFER()
{
HDC_of_MainWindow = GetDC( Handle_of_MainWindow );
Canvas_for_FRONT_BUFFER = CreateCompatibleBitmap( HDC_of_MainWindow, FRONT_BUFFER_WIDTH, FRONT_BUFFER_HEIGHT );
if (Canvas_for_FRONT_BUFFER == NULL)
{
MessageBox(nullptr,L"failed to create Canvas_for_FRONT_BUFFER",L"Error!",MB_ICONEXCLAMATION | MB_OK);
}
HDC_of_FRONT_BUFFER = CreateCompatibleDC( HDC_of_MainWindow );
if (HDC_of_FRONT_BUFFER == NULL)
{
MessageBox(nullptr,L"failed to create the HDC_of_FRONT_BUFFER",L"Error!",MB_ICONEXCLAMATION | MB_OK);
}
HBITMAP Temporary_HBITMAP = (HBITMAP)SelectObject( HDC_of_FRONT_BUFFER, Canvas_for_FRONT_BUFFER );
DeleteObject ( Temporary_HBITMAP );
ReleaseDC ( Handle_of_MainWindow, HDC_of_MainWindow );
}
I use the following
void Draw_Something_To_FRONT_BUFFER()
{
int W=FRONT_BUFFER_WIDTH;
int H=FRONT_BUFFER_HEIGHT;
StretchBlt(HDC_of_FRONT_BUFFER, 0, 0, W, H, HDC_of_Something, 0, 0, W, H, SRCCOPY);
}
Then I put it all to the screen with
void Draw_From_FRONT_BUFFER_To_MainWindow()
{
HDC_of_MainWindow = GetDC( Handle_of_MainWindow ) ;
BitBlt
(
HDC_of_MainWindow, 0, 0, FRONT_BUFFER_WIDTH, FRONT_BUFFER_HEIGHT, HDC_of_FRONT_BUFFER, 0, 0, SRCCOPY );
ReleaseDC( Handle_of_MainWindow, HDC_of_MainWindow ) ;
}
I want the listbox to be added to the front buffer (this is not the screen). It is a collection of buffers that I am combining then later placing them to the screen.
Why does the following not work? It works without the ShowWindow part added. It does not show the listbox this way. What should I be doing to put the listbox in the buffer? I do not want to put a final buffer to the screen and then (after that) put the listbox to the screen. I want it all in the buffer and then the buffer alone to be put to the screen.
void Draw_Something_To_FRONT_BUFFER()
{
int W=FRONT_BUFFER_WIDTH;
int H=FRONT_BUFFER_HEIGHT;
StretchBlt(HDC_of_FRONT_BUFFER, 0, 0, W, H, HDC_of_Something, 0, 0, W, H, SRCCOPY);
ShowWindow(Handle_of_ListBox, SW_SHOW);
}
How do I add listboxes to a back buffer and use that?
Thanks.
|
|
|
|
|
Painting information to the screen should be in the WM_PAINT handler. If you do it in other parts of the program then you will not always get the correct information in the display.
|
|
|
|
|
Do you mean like this?
case WM_PAINT:
{
PAINTSTRUCT ps;
HDC_of_MainWindow = BeginPaint(hwnd, &ps);
Draw_From_FRONT_BUFFER_To_MainWindow();
ShowWindow(Handle_of_ListBox, SW_SHOW);
EndPaint(hwnd, &ps);
return 0;
}
break;
I tried that already.
The listbox seems to be there but invisible.
The way that I did it before, the listbox was there but invisible.
I could detect clicking on the area where the listbox was supposed to be was not allowing me to click on the main window.
I click where the listbox is blocking other things, but nothing happens.
It is invisible or just blocking my clicks to the main window. I want to see the listbox and interact with it.
|
|
|
|
|
A ListBox is just a Window, so if you call ShowWindow using its handle, then it may cover part of your main window. That in turn will receive a WM_PAINT message and redraw itself, which may well have the effect of hiding the ListBox. I am really having difficulty understanding what you are trying to do here. The most common use of ListBoxes is as part of a DialogBox, or CFormView. If you want it as part of your MainWindow then you need to add it into the client zone, and not overwrite it with the main window.
|
|
|
|
|
Before I go any further. Thank you for your help.
I am using a listbox just as an example.
It is the window (a window) itself which is represented by a listbox that I am interested in being able to place in a buffer and then placing that buffer to the screen.
It could be a textbox (created via a window) or some subclassed window from elsewhere like a subclassed browser window or a subclassed game, etc. I want to be able to put all of this in a buffer before placing the result to the screen. Does that help to explain it?
Here is where this issue was discovered:
I noticed that sometimes on some computers that multiple items show up on the screen one at a time slow enough that I and other people detect a lag in the screen being fully created. Memory overloaded, or cpu overloaded, or too much other programs running for smooth response at the screen level. I do not want that. I prefer that my program places everything in a buffer then places that one cohesive buffer to the screen, and not one piece at a time. I can do this with text. I can do this with bitmaps. I should be able to do this with windows (listboxes, textboxes, subclassed outside windows).
I tried. I studied. I tried placing almost randomly attempts to break through my ignorance of how to do this.
How to do it?
|
|
|
|
|
PotatoSoup wrote: I noticed that sometimes on some computers that multiple items show up on the screen one at a time slow enough that I and other people detect a lag in the screen being fully created. Memory overloaded, or cpu overloaded, or too much other programs running for smooth response at the screen level.
First of all remove the calls such as
ShowWindow(Handle_of_ListBox, ...);
from the WM_PAINT message handlers.
|
|
|
|
|
OK.
I put things back they way they were.
Now i have tried using
SetFocus(Handle_of_ListBox);
But that did not work.
I can tell that something is still there because I can not click on that area of the main window.
|
|
|
|
|
PotatoSoup wrote: Now i have tried using
SetFocus(Handle_of_ListBox);
But that did not work.
I'll ask you again: define "did not work." 
|
|
|
|
|
I guess that meant that I was disappointed with my current attempt to add a generic window to a buffer and then to be able to put that buffer to the screen (showing the window).
I have been trying to create or subclass a window (in this example a listbox), and place that window (I guess as a child window?) into or on a buffer, and then place the buffer to the screen. It "did not work." meant that I failed in the attempt.
Same goal of adding a listbox to a back buffer and using that. Generally, the listbox is just an example of some window that I can get a handle of in my attempts.
"did not work." = "I failed at this again."
|
|
|
|
|
I am still suggesting you are doing this all wrong but I can fix this problem if you really are manually drawing the list.
To me it sounds like the listbox should be a child within some other window and you are just trying to do this in some crazy manual way. If it was child it would draw itself when needed with zero code needed from you. However if you are manually doing it you could try this
Copy all the code in the WM_PAINT function of your listbox and place it inside a function like so
void PaintMyList (HWND Handle_of_ListBox, HDC Dc)
{
}
Now replace the PaintStruct stuff and simply us the HDC passed on the interface and use Handle_of_ListBox
where you need a handle .. okay check it all compiles. So we are clear all you are using from the PaintStruct
is the DC and the DC you are to use is declared on that function interface so you can remove all references
to the PaintStruct.
Now use a variant you posted above with the one call change
case WM_PAINT:
{
PAINTSTRUCT ps;
HDC_of_MainWindow = BeginPaint(hwnd, &ps);
Draw_From_FRONT_BUFFER_To_MainWindow();
PaintMyList(Handle_of_ListBox, ps.hdc);
EndPaint(hwnd, &ps);
return 0;
}
break;
In vino veritas
modified 7-Apr-20 3:30am.
|
|
|
|
|
This approach, even if you somehow managed to get it to work, sounds kludgy at best.
If your program is causing this much lag, it sounds as though way too much work is going on in the UI thread. The main/primary thread should do little else besides UI-related stuff. Everything else should be handled by secondary threads.
"One man's wage rise is another man's price increase." - Harold Wilson
"Fireproof doesn't mean the fire will never come. It means when the fire comes that you will be able to withstand it." - Michael Simmons
"You can easily judge the character of a man by how he treats those who can do nothing for him." - James D. Miles
|
|
|
|
|
"sounds kludgy" is a start. I can accept that as the beginning of a suggested answer.
I can multi-thread. If you are saying that you know of how to solve this by multi-threading and thus placing a window that is (in this example) a listbox (or some other generic window) to a back buffer, I am ready for that instruction. If you simply did not read my original question and then fired off a general cutting response with no valid logic, then I can understand that and it is ok. I have scanned and responded wrongly myself a few times. If you have a valid solution keeping to answering my first post, then I am ready to read it. If you can this with multi-threading, then please tell me how.
I found this lag in someone else's program that I had nothing to do with coding. I do not recall seeing it in any of my current programs that I wrote. Their program is a really nicely written program that does a lot of things well. But, when their program is running and at the same time some other huge programs (plural) with intense cpu use and intense memory use are running, then I saw a lag that was enough for me to see postings of sections of their program to the screen.
If someone is running my programs then however intense the cpu and memory usage is by any other running programs, I do not want the user to see any lag of sections of my main window being put to the screen. I would prefer that my programs get its work done building a screen buffer before placing anything to the screen.
Thank you for asking.
|
|
|
|
|
Perhaps SetThreadPriority if you're concerned about competing with CPU pigs?
|
|
|
|
|
You guys are great. You seem to have misread what I am asking and yet you still gave an answer that I find useful. Thanks. I could use your suggestion of SetThreadPriority for my screen capture of the listbox (in using a dedicated thread for the capture) if I do not find a way to place the listbox into a buffer. I like this site.
I still would like to be able to put a listbox (or other window) into a graphics buffer.
|
|
|
|
|
I'm sorry, but that's all I can offer. I write pure C++ with a thin adapter for Windows, so I've never done MFC things. I just tossed it out there because it seemed to address your most recent post.
|
|
|
|
|
"I write pure C++ with a thin adapter for Windows". That is one of the most impressive comments that I have read on this and most programming sites within the last year. That reads like an advisable goal.
How would you approach a quest like this as my first post explains?
Thank you.
|
|
|
|
|
Victor states that you're not doing anything with MFC, so I must have jumped to the wrong conclusion when I saw your code creating a window. I just run my stuff as a console app that ultimately uses nothing but std::cin, although it can also read commands from a file.
|
|
|
|
|
The OP's code has nothing to do with MFC. It is a plain Win32 code.
|
|
|
|
|
Does someone here know how to put a text box or a list box to a back buffer? Should he have looked more into blitting it there? The discussion is valuable. I liked reading it and learning from it, but I think that he still has the problem of putting the list or text box to a buffer.
Please someone help this guy and I will try to learn from this also.
Thank you.
|
|
|
|
|
PotatoSoup wrote: How to do it?
Simple, make the ListBox a child of the main window and size it to the client area. That is the standard Windows paradigm that has worked since the 1990s. No need for complicated backdoor processing or multiple device contexts. The ListBox will happily repaint itself whenever necessary.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
