|
Hi Don,
I am using fread to read contents from a file and the return value is size_t. Then I need to pass the size_t value to another function which accepts unsigned int as input parameter, and this is why I need to do the type conversion.
Any suggestions to solve this issue? (which types should I convert to? Any possible lost of data?)
regards,
George
|
|
|
|
|
Pragma warning disable! It doesn't hurt when you send a size_t to an int param. You could do like (int)size_tVariableHere, when you pass size_t as an int argument. The "possible loss of data" could be about the sign information. size_t is unsigned, We can be sure size can't be in negative, so you can very well use it. It's not a problem at all.
btw : add this in your header:
pragma warning( disable : C4267 )
He's become a household word in the Lounge. A whole new phraseology has evolved. Post a link or reply with a smiley and rose, and you've made a "Satipsism". So what? It's an interesting thing about the Internet, the evolution (as in change, not progress) of tone, quality, terminology, etc.
-Marc Clifton.
Best wishes to Rexx[^
|
|
|
|
|
Hi VuNic,
I find even if I convert to an unsigned int, the warning message is the same. I am using Visual Studio 2003.
Do you know the reason?
regards,
George
|
|
|
|
|
I would like to tell that size_t is an unsigned int. If u convert it into int, then it may have loss of data. But as u r saying, u r sending to a function accepting unsigned int, then there should not be any warning like such.
But as u r passing the return value of fread and u won't read more than 2147483647 bytes at a time, it won't have any problem. Let the warning appear.
Come online at:-
jubinc@skype
|
|
|
|
|
Thanks Don,
I am using Visual Studio 2003. I think there should not be any warning if I convert size_t to unsigned int, but there actually a warning. Strange.
regards,
George
|
|
|
|
|
I know this was already answered but you are obviously not trying to covert it to unsigned int.. You are correct that a size_t is an unsigned int, so technically no conversion is needed. You are trying to convert an unsigned int to an it, which could result in a loss of data if the value exceeds 32,768 (3K).
In any case the ‘warning…’ means exactly what it says.
INTP
"Program testing can be used to show the presence of bugs, but never to show their absence."Edsger Dijkstra
|
|
|
|
|
Hi John,
I do not know why there will be data lost if I convert size_t to unsigned int. Could you show me an example please?
(I am using 32-bit machine.)
regards,
George
|
|
|
|
|
You had me worried there for a moment – I thought I did not make it clear. No conversion is needed, the warning says you are converting it to an ‘int’ which will result in possible data loss – if and only if the value exceeds 2,147,483,647 (corrected for 32-bit integer).
The warning is not saying that you have a possible loss of data when converting to ‘unsigned int’, it says you are converting it too ‘int’, which is why the warning is occurring.
INTP
"Program testing can be used to show the presence of bugs, but never to show their absence."Edsger Dijkstra
|
|
|
|
|
Hi buddy,
You can try that even if we convert to unsigned int, the warning message is the same in Visual Studio 2003.
Any ideas?
regards,
George
|
|
|
|
|
 You are still talking about the warnings? These warnings are meant to be ignored once you know what they are warning about. I've a piece of code that uses STL Map, and when you compile, the compiler takes minutes to stop it's warning process. Where you should only use the pragma-warning-disable.
He's become a household word in the Lounge. A whole new phraseology has evolved. Post a link or reply with a smiley and rose, and you've made a "Satipsism". So what? It's an interesting thing about the Internet, the evolution (as in change, not progress) of tone, quality, terminology, etc.
-Marc Clifton.
Best wishes to Rexx[^
|
|
|
|
|
Thanks VuNic,
Do you think convert from size_t to unsigned int will cause data lost on 32-bit platform?
regards,
George
|
|
|
|
|
George_George wrote: Do you think convert from size_t to unsigned int will cause data lost on 32-bit platform?
You will loose if you send signed number. Ok, test it by yourself :
<br />
int nIntNeg = -100;<br />
int nIntPos = 100;<br />
size_t sz;<br />
sz=nIntPos;<br />
cout<<"\n"<<sz;<br />
<br />
sz=nIntNeg;<br />
cout<<"\n"<<sz<<endl;<br />
The point here is not about you will/will-not loose data, but it's to say that you wont get into a situation where a function would return signed number for a size_t type.
Eg:
<br />
std::string st;<br />
st="";<br />
cout<<st.size();<br />
At the worst, st's size can me only 0, can't go below that. So you get the point? size_t s are meant for unsigned data. If you send signed numbers, you WILL loose it.
He's become a household word in the Lounge. A whole new phraseology has evolved. Post a link or reply with a smiley and rose, and you've made a "Satipsism". So what? It's an interesting thing about the Internet, the evolution (as in change, not progress) of tone, quality, terminology, etc.
-Marc Clifton.
Best wishes to Rexx[^
|
|
|
|
|
Hi VuNic,
I can understand your points. But I am always using unsigned int. Do you think there will be any data lost if I convert size_t to unsigned int on 32-bit machine?
regards,
George
|
|
|
|
|
|
Hi Roger,
I am using 32-bit only (no 64-bit). I have read the URL from MSDN and there may be warnings when we are using 64-bit platform. But I do not know why in my 32-bit environment, there is still the warning message.
regards,
George
|
|
|
|
|
pragma warning( disable : C4267 )
Did you do this or not?, put that in the afx.h header.
He's become a household word in the Lounge. A whole new phraseology has evolved. Post a link or reply with a smiley and rose, and you've made a "Satipsism". So what? It's an interesting thing about the Internet, the evolution (as in change, not progress) of tone, quality, terminology, etc.
-Marc Clifton.
Best wishes to Rexx[^
|
|
|
|
|
Hi VuNic,
You mean it is a false warning?
regards,
George
|
|
|
|
|
Here[^]
Relatively it's a simple question, wondering it required so much explantion. Think it this way,
Floats are meant for storing percision data, if you store a float value in an int? wont it loose the precision? But you are not going to do that right? the same way unless or otherwise you manually do something like :
<br />
void myFunction(size_t sz)<br />
{<br />
<br />
}<br />
and pass something like
myFunction(-100);
There's no possiblity of loosing data. In short, It looks you aren't sending negative values in your code so please ignore the warning.
He's become a household word in the Lounge. A whole new phraseology has evolved. Post a link or reply with a smiley and rose, and you've made a "Satipsism". So what? It's an interesting thing about the Internet, the evolution (as in change, not progress) of tone, quality, terminology, etc.
-Marc Clifton.
Best wishes to Rexx[^
|
|
|
|
|
Hi VuNic,
I am always using unsigned int as data type. Example,
unsigned int i = <some value="">;
size_t new_value = (size_t)i;
I am not sure whether there will be any data lost if I convert i to new_value as size_t type. I can understand convert to signed int will cause data lost. What about unsigned int?
Looks like we are talking about two different things. 
regards,
George
|
|
|
|
|
if your sure you will not go between 32-bit and 64-bit with this code someday then an explicit cast should eliminate the warning. Let the compiler know the conversion is not implicit and it will take your word for it that you know what your doing.
size_t First=45;
int Second=67;
unsigned int Third=98;
// Damned if you do
Second=(int)First;
Third=(unsigned int)First;
// Damned if you don't
Second=static_cast<int>(First);
Third=static_cast<int>(First);
|
|
|
|
|
Hi bob16972,
I am using C and I can not use static_cast template class. I am wondering if I am building application for 32-bit platform only, will there be any data lost if I convert size_t to unsigned int? (not signed int)
regards,
George
|
|
|
|
|
The static_cast was in my post to appease the C++ purists. I use the C-style casts myself.
George_George wrote: I am building application for 32-bit platform only, will there be any data lost if I convert size_t to unsigned int?
No.
Clarification:
I included signed int in my original post without thinking as there is the potential to misinterpret or lose data with that conversion. size_t is supposedly an unsigned int so the conversion from size_t should be pretty clean in a 32-bit only world.
|
|
|
|
|
Hi bob16972,
static_cast is an operator in C++, how do you think we can use it in pure C program?
regards,
George
|
|
|
|
|
Considering this is a C++ forum, I assumed you were using .cpp files. If you are truly using .c files then my apologies. I haven't compiled a "pure" C program since 1992 or 1993.
Either way, I don't use static_cast myself so no harm no foul.
|
|
|
|
|
Thanks bob16972,
It is ok.
regards,
George
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
