|
Hey all,
My last [meaningful] programming experience was programming assembly on the 6502/6510 and the 6802 MPUs in the early eighties. I'm tending to relate the function of pointers in C++ to the role of the "Indexed Addressing Mode" in assembly (e.g. LDA $0400,X), but there seems to be a significant layer of abstraction that I'm not grasping. I've never programmed in C or C++ or really any HLL since the 80's and I'm having tremendous difficulty with pointers.
I see a lot about how great and useful they are, but aside from the mundane examples, their real usefulness evades me. The examples likely containing *practical* examples are equally likely far beyond my C++ proficiency level. Given this relative impass, can someone explain in general terms what application might be aided by pointers? (Indeed, there must be some application in which pointers are absolutely necessary.)
Naturally, links to existing resources would be great as well.
I'm certain that relating pointers with indexed addressing is a significant fallacy, but right now it's all I have to work with.
Thanks,
fadibitz
If you see only one solution, then you don't understand the problem.
|
|
|
|
|
Here are a couple reasons to use pointers:
1) Dynamically allocating memory.
Once you dynamically create a type, you need a pointer to locate it and employ it. Also, dynamically a type or a collection of types, makes your applcation more scalable.
2) Passing large objects or collections by reference.
Passing a large object to a function by value means you have to copy the object and waste time and memory. If you pass it by reference via pointers, only the pointer gets copied.
"We make a living by what we get, we make a life by what we give." --Winston Churchill
|
|
|
|
|
In addition, sorting is much faster by organizing the pointers rather than moving the data around.
Searching can be made much faster using pointers in a specific manner as well.
There are a number of reasons to use pointers over other data types, but only the developer (such as yourself) can determine if the pointers are truly necessary.
Phil
|
|
|
|
|
All of these responses make an icredible amount of sense. Thanks for the wide view.
It's looking like the way to procede is to take my "overview" of pointers and move on to other programming topics like structures, classes, and templates. My guess is that as I progress through these programming concepts, techniques and examples will call on ever more advanced usages of previously-learned things. Pointers included. Trying to understand them fully without benefit of prior experience or real understanding of c++ fundamentals is (or seems) destructive and self-defeating.
What say you all?
Robert Nicholas
---------------
If you see only one solution, then you don't understand the problem.
|
|
|
|
|
My first suggestion is for you to "play" or "Experiment" with pointers until you understand what you can and should not do.
Phil
|
|
|
|
|
Coming from assembly before C/C++ myself - pointers are the closest to a one-to-one relationship
with CPU addressing. It's the lowest level abstraction, just above "Indexed Addressing Mode" in
assembly. The benefit is performance and control. The drawback is when mistakes are made coding,
they are bad - the compiler can't save you  Higher level languages hide these issues,
sacrificing a bit of performance for safety and ease of use (or hiding the concept of memory
addressing altogether).
My 2 cents.
Mark
"Posting a VB.NET question in the C++ forum will end in tears." Chris Maunder
|
|
|
|
|
Yes, power corrupts. When you mess up, it corrupts absolutely.
"We make a living by what we get, we make a life by what we give." --Winston Churchill
|
|
|
|
|
Confirmed!
Robert Nicholas
---------------
If you see only one solution, then you don't understand the problem.
|
|
|
|
|
You have received some good answers so I will keep this simple.
A pointer is the address to a block of memory, just as a street address is the address of the house and not the house itself. If you wish to have a separate function manipulate the contents of the memory block, then you need to pass the address to that function so that it knows where the memory block is located.
The C string manipulation functions are good examples of using pointers: see strcpy amongst others.
If you want to create your own liked list, instead of using the STL ‘list’ template, then pointers are essential: struct node { some data variables; struct node* pNextNode; };
Example (in generic C):
void clear_buffer(void* pBuffer, unsigned nSizeInBytes)
{
unsigned char* p = (unsigned char*)pBuffer;
unsigned i;
for( i = 0; i < nSizeInBytes; ++i )
p[i] = 0;
}
INTP
"Program testing can be used to show the presence of bugs, but never to show their absence."Edsger Dijkstra
|
|
|
|
|
Hi,
I use the function ShellExecute to start an application, say MyApp.exe. Does anyone know how to terminate the process of MyApp?
It seems ShellExecute can return a handle (the first parameter of ShellExecute). I tried to use TerminateProcess to end MyApp but I failed. It seems the handle is not the "real handle" of the process, right?
Thank you a lot for your help!
Jeffrey
|
|
|
|
|
I've never tried to terminate something spawned from ShellExecute so I can't help there but...
A quick search of my projects, I found one that uses CreateProcessWithLogonW and later if problems develop it resorts to using TerminateThread utilizing the handle in the processInformation structure that is filled in CreateProcessWithLogonW. CreateProcess has a similar parameter list and utilizes the processInformation as well so I would guess you should be able to use CreateProcess for the same purpose though I've never tried it.
Hope that helps.
|
|
|
|
|
You'll need to use ShellExecuteEx() to get a handle on the new process.
|
|
|
|
|
Thank you, Bob and Mike.
When I get the handle on the new process, do I call TerminateProcess to close it? Thank you, Mike.
Jeffrey
|
|
|
|
|
sunshine jeffrey wrote: When I get the handle on the new process, do I call TerminateProcess to close it?
Yep
|
|
|
|
|
Although sometimes the only way, in general I question the wisdom of using TerminateProcess to stop a process. Doing so could result in corruption of data files for example. It would be better to try to close down the process gracefully first and only terminate as a last resort. The WM_CLOSE message and the GenerateConsoleCtrlEvent function are of interest in this respect.
Steve
|
|
|
|
|
I agree with Stephen. When using TerminateProcess, you have no idea what the process is doing at
that instant or what state it's in. If at all possible, there sould be some kind of interprocess
communication used to let the spawned process close itself gracefully.
Mark
"Posting a VB.NET question in the C++ forum will end in tears." Chris Maunder
|
|
|
|
|
Thank you for your suggestions, Stephen and Mark. I also kinda feel TerminateProcess a little brutal  I will try the ways as suggested. Thanks
Jeffrey
|
|
|
|
|
What is message loop in MFC
|
|
|
|
|
|
|
hello,
I have a crystal report (version 8.5) which has about 10 on-demand sub-reports embedded in it. But when the report opens, the hyperlinks for them do not work, as in, they appear as plain-text, not clikable.
The same report works just fine when viewed in crystal software. I am using visual c++ 6.0. Is there any addtional coding to be done to enable it.?
Also, how do I determine which subreport the user has clicked on, in order to open the appropriate sub-report. Right now, all i am able to do is to scroll through each of these subreports. But i dont know how to stop and display the one that the user has clicked on.
Kindly help
|
|
|
|
|
I would like to know whether a file(*.txt) is normal ansi text file or unicode file by program. Is there standard class in C++/MFC for reading UNICODE file?
Best Regards,
Suman
|
|
|
|
|
As with anything else, it is not a straight forward process:
http://codesnipers.com/?q=node/68[^]
"We make a living by what we get, we make a life by what we give." --Winston Churchill
|
|
|
|
|
If the app expects the files to be relatively small, you can read in the file raw into a buffer and use IsTextUnicode() against the buffer using the IS_TEXT_UNICODE_SIGNATURE flag.
In an MFC app, this should occur in the serialize method or be a function called from the serialize method. to get the filesize, get the file pointer using ar.GetFile() and call GetLenth() on the file pointer. Obviously, you'll read in the raw data using ar.Read().
Not the most efficient way I'll bet, but it seems to work. However, note the disclaimer in MSDN in the remarks section for IsTextUnicode()
|
|
|
|
|
AFAIK, there is a two byte signature at the very start of a UNICODE text file.
The signature consists of the character values FF hex (255 decimal) and FE hex (254 decimal).
They can be in the order FEFF or FFFE, which flags if the bytes in the file is big or little endian. I don't remember which is which, but Google is your friend.
You can try it yourself, just create a text file with Notepad, save it in UNICODE and then dump the file with a hexdump (or hexeditor) program.
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
