Re: cin/streambuf-problem

["R. Sinoradzki" <sinoradz at student dot uni-kl dot de>]

Hi,

After reading a bit and digging in the STL I must admit,
that I perhaps misunderstood something.

The following method solves my problem:

cin.ignore( std::numeric_limits<std::streamsize>::max() ,'\n');

I thought a bit different of the underflow()-method in streambuf
and fstream.

I thought, that if an underflow occurs, the streambuffer is filled
to it's maximum size or at least with the data that is available,
if this is less.
In this case in_avail() defined as (egptr() - gptr()) would return a
bigger value.

Regards,
Ralf


"R. Sinoradzki" wrote:
> 
> Hi,
> 
> I have this little program. It does not behave as
> expected with gcc-3.0 and it's stdlib.
> If I type for example "hello" the loop runs 5 times,
> before the next 'cin' because 'cin.rdbuf()->in_avail()'
> reports only 1.
> 
> With gcc-2.95.4 cin.rdbuf()->in_avail() returns 6
> and it works as expected.
> 
> My gcc-3.0pre is an about 10 days old CVS-version. Perhaps
> this is a well-known problem.
> 
> bye, Ralf
> 
> /////////////////////////////////////////////////////////////
> 
> #include <fstream>
> #include <iostream>
> 
> using namespace std;
> 
> int main(){
>     int input=0;
>     do{
>         if (cin.fail()){
>             cin.clear();
>             cin.ignore(cin.rdbuf()->in_avail(),'\n'); // seems to bug
>             cout<<"bad input." << endl;
>         }
>         cout<<"enter a number: ";
>         cin>>input;
>         cout << input << endl;
>     }
>     while(input!=3);
>     return 0;
> }
> 
> //////////////////////////////////////////////////////////