|
The efficiency of the two does not realy differ very much. Of course, the first way comes with some additional machine instructions. It mainly depends on what you want to do with your constructor.
William
|
|
|
|
|
namaskaaram wrote: which one of the above is better??(in terms of efficiency)
I think first one is good to direct the reference correctly. As well we can give the gurantee of consistency in data.
Knock out 't' from can't,
You can if you think you can
|
|
|
|
|
Actually "copy constructor" is this one: C(const C & );
Regarding overloading assignment operators, I think your first variant, which returns a reference, allows multiple assignments like this: a = b = c, or compact expressions like this: f(a = b). The second one does not allow this, but is little faster because of missing return value.
-- modified at 4:24 Tuesday 20th June, 2006
|
|
|
|
|
actually i meant assignment operator overloading....sorry abt the typo! ;o)
thank u all!!!! 
|
|
|
|
|
Viorel. wrote: The second one does not allow this, but is little faster because of missing return value.
You'll save 1, or maybe 2 CPU cycles. (Depending on how the compiler allocates registers, you may end up saving no cycles)
|
|
|
|
|
in fact, prefer the following operator :
C& C::operator = (const C&);
the reason to prefer this one instead of the void operator is that the user of your class will then be able to cascade the assignment operations. consider this :
int i1, i2, i3;
C c1, c2, c3;
i1 = i2 = i3 = 5;
c1 = c2 = c3 = ;
this is possible only because the operator = returns a reference to the object it has just modified...
TOXCCT >>> GEII power
[VisualCalc 3.0 updated ][Flags Beginner's Guide new! ]
|
|
|
|
|
Yes, you're right. The rationale is that the assignment operator should behave in the same way as assignment for primitive types.
Kevin
|
|
|
|
|
namaskaaram wrote: i believe copy constructor can be written in two wayz:
That is not a copy constructor, it's an assignment operator.
|
|
|
|
|
Hi,
lets say that i have an operation that is performed on two sets of data independently,
that is the data is separable, and not related.
so, for an example;
<br />
void Function1(int *a, int *b, int *c, int count)<br />
{<br />
for(int i=0; i< count; i++)<br />
c[i] = a[i] * b[i];<br />
}<br />
I can measure the execution time using QueryPerformanceCounter.
<br />
main()<br />
{<br />
Function1(a,b,c,count);<br />
<br />
OtherFunction(...);<br />
}
My question is, now, if i want to break the function to two threads, to take advantage
of dual cpu, or dual core, how do i
1- Measure the performance of the code..
2- Determine the function has finished, or terminated
Example of an implementation that i am looking at is;
void Function2(int *a, int *b, int *c, int count)<br />
{<br />
AfxBeginThread1()
AfxBeginThread2()
}
<br />
main()<br />
{<br />
Function2(a,b,c,count);<br />
<br />
<br />
<br />
OtherFunction(...);<br />
}
Can anyone help on this?
Thanks
|
|
|
|
|
See Here[^] maybe it is some helpful to you
whitesky
|
|
|
|
|
Hi,
nice link, there's some info on how to wait for a thread to finish, but how do i time the threads?
|
|
|
|
|
Usman Ullah Sheikh wrote: // How do i block OtherFunction from executing before Function2 finishes.
Use criticalSection
<marquee scrollamount="1" scrolldelay="1" direction="up" height="10" step="1">--[![Jig | [Dance]](https://www.codeproject.com/script/Forums/Images/jig.gif) ]--
[My Current Status]
|
|
|
|
|
Usman Ullah Sheikh wrote: // How to measure the time Function2 takes? Function2 is multithreaded.
// Time it takes for both thread to finish,
Why not you use global variables? and update it with the last process that finishes its task?
<br />
void Function2(int *a, int *b, int *c, int count)<br />
{<br />
AfxBeginThread1()
now, if the process is completed, inside the function that thread1 runs,
do your///QueryPerformanceCounter - end
AfxBeginThread2()// for data in the range of count/2 -> count
now, if the process is completed, inside the function that thread1 runs,
do your///QueryPerformanceCounter - end
}
while updating your "end" check if it has been updated by the previous process, if so, one process is already over so no need to worry, the last one you update is the resultant one. If the "end" counter is still 0, no process is yet over. So after updating it, you still need to wait. When the next process gets over it again checks if the "end" counter is 0 or not. This time it finds value inside it (since it's been updated by the previous thread) now you can be sure everything's over and 'end'- this 'start' this is the total time taken.
<marquee scrollamount="1" scrolldelay="1" direction="up" height="10" step="1">--[ ]--
[My Current Status]
|
|
|
|
|
I think that will work! Thanks! I'll try it!
|
|
|
|
|
Surely, the Dialog will close when the Enter and ESC key down.
I know reload the PreTranslateMessage(MSG* pMsg) function can avoid the problem.
In my project, there are some other controls in the dialog, and I want the Edit controls and buttons can get the Enter/ESC command but not close the Dialog, is there any way?
I tried the OnChar(), OnCommad(), there din't work at all.
BOOL CEditDlg::PreTranslateMessage(MSG* pMsg)
{
// TODO: Add your specialized code here and/or call the base class
if(pMsg->message==WM_KEYDOWN)
{
if(pMsg->wParam==VK_RETURN || pMsg->wParam==VK_ESCAPE)
{
if (!GetFocus()->IsKindOf(RUNTIME_CLASS(CEdit)))
return true;// pMsg->wParam=NULL ;
else
{
...............
}
}
}
return CDialog::PreTranslateMessage(pMsg);
}
|
|
|
|
|
if your problem is in editbox do you have ES_WANTRETURN in property edit
whitesky
|
|
|
|
|
My edit controls are single-line.
in fact, what I want is send a message to Parent window, when the focused edits/buttons get Enter/ESC key.
|
|
|
|
|
Try WM_GETDLGCODE
Regards,
FarPointer
Blog:http://farpointer.blogspot.com/
|
|
|
|
|
That's why PreTranslateMessage is not a very clean solution. Prefer the solution described in the FAQ[^]. In this way, you still have the complete flow of messages inaltered, it is just the closing of the dialog that is suppressed and you don't need to make tricks to get back your esc and enter messages when you need them.
Cédric Moonen
Software developer
Charting control
|
|
|
|
|
then, how to handle the enter/esc key for Edit control and buttons?
And seems that it doesn't enter into the OnChar() at all.
|
|
|
|
|
zeus_master wrote: And seems that it doesn't enter into the OnChar() at all.
Then your problem is elsewhere, it has nothing to do with the destruction of the dialog on enter or Esc.
Cédric Moonen
Software developer
Charting control
|
|
|
|
|
yes
the code below <b>else </b> is where I need your kindly help.
if (!GetFocus()->IsKindOf(RUNTIME_CLASS(CEdit)))
return true;// do nothing and ignore the key process
else
{
<u><b> ...............
//add proper code for EDIT control capture the Enter and ESC key.</b></u>
}
|
|
|
|
|
Why are you doing that in PreTranslateMessage ? I think you can simply add a message handler for WM_CHAR or WM_KEYDOWN for the edit control no ? I'm not sure about that, so you have to check.
Cédric Moonen
Software developer
Charting control
|
|
|
|
|
I tried, and failed......
if there is no better methord,I would have to add the edit proc function in PreTranslateMessage() directly...
|
|
|
|
|
zeus_master wrote: if (!GetFocus()->IsKindOf(RUNTIME_CLASS(CEdit)))
return true;
Your logic seems to be wrong. You want the condition to be true right?
[modified]
By returning true from PreTranslateMessage(...), you are telling your application to bypass any default handling of the message.
[modified]
Last modified: Tuesday, June 20, 2006 3:28:49 AM --
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
