|
I was hoping for the opposite effect
making functions not inline that usually would Sorta'
class MyClass{
theFunc(){ AfxMessageBox("hey"); }
theFunc2()
{
AfxMessageBox("I don't want this function to INLINE");
m_bFake = FALSE;
}
};
Thanx anyways
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
There's a setting in the C++ tab in the optimizations category that allows three settings - disable, only inline those explicitly marked 'inline' (believe the cl'er can pass final judgement here, but should warn), and 'hey Ms compiler, you're so good at this stuff, why don't you decide what to inline'.
Would that help - or is it really a template related issue - post some code?
|
|
|
|
|
As Jimi and Joaquín M López Muñoz say: you can do nothing except declare them correctly and let the compiler do its job according to your wishes (read /options).
Besides of this you should ask yourself another question: if the compiler thinks that a function may be inlined, why you think it should not?
Dont tell me that the function may be too big. The argument "the resulting code will be too big" is nonsense nowadays. Applications can expect "sufficient" memory to run almost any code.
So? What is left? Speed? If it gets called so many times, the compiler is right - it preserves lot of overhead for the function call and execution.
Anything else that I missed?
|
|
|
|
|
Andreas Saurwein wrote:
Besides of this you should ask yourself another question: if the compiler thinks that a function may be inlined, why you think it should not?
I always thought it was good practice to only inline those functions which were relatively small(a few push, mov and pop, ret)
My question was that i'm aware the compiler will implicity inline functions defined and declared in the class itself, but only under certain circumstances(NOT recursion, etc...)
Is there a way i can avoid this gauranteed!!!
3 member functions dec/defined in the class itself I feel are far to large to inline.
1) 500 bytes total( 51 prolog, 26 epilog) so i'd save 77 bytes if I inlined, but I'd have additional 423 bytes * already 9 function calls = 3807 bytes which could be avoided if the function didn't inline.
So under most circumstances I agree it's best to leave the compiler to do it's job, but this is an extenuating circumstance(I think).
I was unsure if the compiler would INLINE this function of mine, because it's a template class and all the functions have to dec/defined within the class(or they won't show in ClassView...? )
I suppose i could have looked at the disassembly and determined if it was calling or inlining??? But i'm too lazy;P
Andreas Saurwein wrote:
Dont tell me that the function may be too big. The argument "the resulting code will be too big" is nonsense nowadays. Applications can expect "sufficient" memory to run almost any code.
Most certainly they can, with virtual memory and all(however i'm running a 266,Win98 32 MB RAM, which I find slow). However I feel this goes against what C/C++ stand for. According to(I think is the original "C Programming Language" By Kernighan&Rithchie) C is a relatively low-level language, so I'll assume C++ just follows suit, except better type checking, implementation hiding, etc...both support pointers and inline _asm
So if i didn't wanna worry about code execution and size I'd probably go back to VB...instead I find myself leaning more towards assembly 
Have a day!
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
You can use #pragma auto_inline( [{on | off}] ), or compiler options /Ob
I vote pro drink ![Beer | [beer]](https://codeproject.global.ssl.fastly.net/script/Forums/Images/beer.gif)
|
|
|
|
|
The pragma is much preferd
Could I not use function and intrinsic to accomplish the same thing...?
Thanx again
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
Does the following seem appropriate...?
Type* temp = m_pBuff;
try{
m_pBuff = (Type*)HeapReAlloc(m_hHeap, HEAP_ALLOC, m_pBuff, nHeapSize);
}
catch(...){
bFull = TRUE;
m_pBuff = temp;
}
Any suggestions?
TIA
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
As it stands, HeapReAlloc does not throw any exception (it returns NULL on failure), so your exception handler will never be called. You can add the flag HEAP_GENERATE_EXCEPTIONS to indicate HeapReAlloc to throw on exception if it fails, but even so the technique is probably more expensive (in computational terms) than the simpler check against NULL (IMHO).
Joaquín M López Muñoz
Telefónica, Investigación y Desarrollo
|
|
|
|
|
HEAP_ALLOC macro actually expands
HEAP_NO_SERIALIZE | HEAP_GENERATE_EXCEPTIONS | HEAP_ZERO_MEMORY
sorry bout that.
I can't for some reason get HeapAlloc/ReAlloc to return NULL
even if i try it with 2,000,000,000 byte request it succeeds(without GEN_EXCEPTIONS)...but the exception fires, so i dunno...
Other than this...is everything OK...?
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
 It's weird what you're saying... You might try calling GetExceptionCode in the handler to see what exception is being thrown (STATUS_NO_MEMORY or STATUS_ACCESS_VIOLATION).
Other than this...is everything OK...?
I guess it is, but this is surely masking some underlying problem that probably will strike back in some other form sooner or later. I'd suggest you investigate the reasons why HeapReAlloc seems to behave differently depending on whether it is raising exceptions or not.
Joaquín M López Muñoz
Telefónica, Investigación y Desarrollo
|
|
|
|
|
In an application, I have used a class derived from CHtmlView. I wish to perform the typical find operations on it i.e. find a word and highlight it, find the next occurrence of the word, find the previous occurrence of the word.
Whenever I use the function findText of IHTMLTxtRange, with the parameter "forward " set to a +ve value, the forward search occurs correctly. But if I use the function with a negative value of forward, then the function doesnt work. (although it should,as per the documentation)
WHAT MAY BE THE REASON?
HERE IS A CODE SNIPPET OF WHAT I DID:
IHTMLDocument2 *lpHtmlDocument = NULL;
LPDISPATCH lpDispatch = NULL;
lpDispatch = m_pHelpDataView->GetHtmlDocument();
ASSERT(lpDispatch);
lpDispatch->QueryInterface(IID_IHTMLDocument2, (void**)&lpHtmlDocument);
ASSERT(lpHtmlDocument);
lpDispatch->Release();
IHTMLElement *lpBodyElm;
IHTMLBodyElement *lpBody;
IHTMLTxtRange *lpTxtRange;
lpHtmlDocument->get_body(&lpBodyElm);
ASSERT(lpBodyElm);
lpHtmlDocument->Release();
lpBodyElm->QueryInterface(IID_IHTMLBodyElement,(void**)&lpBody);
ASSERT(lpBody);
lpBodyElm->Release();
lpBody->createTextRange(&lpTxtRange);
ASSERT(lpTxtRange);
lpBody->Release();
CComBSTR search(FindName.GetLength()+1,(LPCTSTR)FindName);
bool bFound,bTest;
long t;
***lpTxtRange->findText((BSTR)search,forward,bMatchCase+bMatchWholeWord,(VARIANT_BOOL*)&bFound);
........
***lpTxtRange->select();
|
|
|
|
|
1)the caret will be destroyed by system? or should be destroyed by programmer?
2)if there is two mdi child windows, one is closed, which message need to be handled to show the caret on the other mdi window? even i don't know how to
process the OnActivateView method.
|
|
|
|
|
DestroyCaret() should be called in the WM_KILLFOCUS by the programmer.
If my memory serves me correctly
WM_SETFOCUS
CreateCaret()
SetCaretPos()
ShowCaret()
WM_PAINT
HideCaret()
WM_KILLFOCUS
DestroyCaret()
HTH
cheers
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
what does the method OnActivateView serve as?
i did as you said, but when i close one view, the other view won't set its caret, why? but if i use another program such as MS VC ++ and then change to it, the caret is here, how to resolve it?
thanks
one is the other who absorbs the thoughts of others
|
|
|
|
|
ooze wrote:
what does the method OnActivateView serve as?
Sorta like an ActiveX event. You override the function to supply custom handling when it's called by the framework.
ooze wrote:
i did as you said, but when i close one view, the other view won't set its caret, why?
Do you have the appropriate code in both view's OnSetFocus/OnKillFocus...?
p.s-In WM_PAINT hide your caret at start and Show at finish.
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
here is the code in OnSetFocus, and onKillFocus
CView::OnSetFocus(pOldWnd);
// TODO: Add your message handler code here
if(CONNECTED)
{
::CreateCaret(this->m_hWnd ,0,ftWidth,ftHeight);
SetCaretPos(CPoint(xCurPos,yCurPos));
::ShowCaret(this->m_hWnd );
}
CView::OnKillFocus(pNewWnd);
// TODO: Add your message handler code here
DestroyCaret();
one is the other who absorbs the thoughts of others
|
|
|
|
|
Code looks fine, however here are some suggestions.
Make sure CONNECTED is true and use all hardcoded values just to make sure everything is prim and proper.
if(TRUE)
{
::CreateCaret(this->m_hWnd, NULL, 1, 20);
SetCaretPos(CPoint(10,10));
ShowCaret();
}
I've had problems with caret functions only to realize variables weren't holding proper values, so make sure
(i'll assume ftWidth is font width)
ftWidth is actually non-zero each call, as with the other vars.
I've never used carets in MDI apps, however I don't see any difference then SDI...
HTH;P
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
all the question now is when i close one MDI View, the other MDI view wont display the Caret, but when it is overlapped or covered by other programs,
it will display it right
why?
thanks for your suggestion
one is the other who absorbs the thoughts of others
|
|
|
|
|
That problem I have had and I can't say forsure why this is
I only found a quick hack and it worked so I went with it...
If the caret appears when you move something over it or resize the window, it's obvious it's a OnPaint not getting called at appropriate times.
So what i did was call the views Invalidate() member
Try this it should work
I guess call it from with the other views OnKillFocus...?
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
I want to write a cd burn or cd maker program which can create the .iso .bin .cue files and burn it to the cdr(w).
But I can't find any reference to this. Could anyone give me an idea?Maybe the format reference? The source code,url are even good.
BTW:The source code I got are under cygwin. I need the true win32. Thanks.
|
|
|
|
|
It's been a while but try visiting the Wotsits file format collection
Wotsits
"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr
|
|
|
|
|
OK, this has probably been asked a 1000 times, so here we go with number 1001. I've been programming C++ for about 5 or so years now, and have yet to find a decent explanation of the various casting operators (reinterpret_cast<>, const_cast<>, etc.). Not to say that several don't exist on here, I just couldn't find them. So, my question is can anybody explain what each of the different C++ cast operators do? Are they part of the language, or do they belong to the STL? Where would I find a good explanation or two here?
Thanks in advance.
Jamie Nordmeyer
Portland, Oregon, USA
|
|
|
|
|
They are part of the core language, and const_cast removes constness, reinterpret_cast allows casting bewtween unrelated types, static_cast does not return NULL if a cast fails, and dynamic_cast does. I'm sure there's a bit more to it than that, but I find myself using dynamic_cast the most, so it's the only one I know really well.
Christian
I have come to clean zee pooollll. - Michael Martin Dec 30, 2001
Picture the daffodil. And while you do that, I'll be over here going through your stuff.
|
|
|
|
|
They are keywords of the language. Short explanation:const_cast<> removes away constness.
static_cast<> is used for conversions that, although well defined in C++, are not automatically applied by the compiler. Most important examples are truncating arithmetic castings (eg. from double to int) and casting from a base class pointer to a derived class pointer.
dynamic_cast<> plays a role similar to that of static_cast<> when downcasting from a base class to a derived one, but it has the important difference that the downcasting is checked for validity at run-time, ie. dynamic_cast<> returns 0 if what you are trying to downcast to say A* is not actually a A*. For this to work, classes passed to dynamic_cast<> must be polymorphic, ie. they must have at least one virtual function.
reinterpret_cast<> is used for anything else, as conversions with no defined meaning in the standard, like for instance converting a pointer to an int.
Joaquín M López Muñoz
Telefónica, Investigación y Desarrollo
|
|
|
|
|
const_cast also removes the volatile attribute.
--Mike--
"Everyone has figured out what 'service pack' really means, so they had to go and change the language. Perhaps this is what Bill was talking about in the 'security is top priority' letter."
-- Daniel Ferguson, 1/31/2002
My really out-of-date homepage
Sonork - 100.10414 AcidHelm
Big fan of Alyson Hannigan.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
