Martin,

Per section 27.6.1.3 of the ANSI C++ draft (in a nutshell):

1) basic_istream<charT,traits>& get(char_type&c);

   effects: if it fails, calls setstate(failbit).

2) basic_istream<charT,traits>& get(char_type* s, streamsize n, char_type
delim = traits::newline());

effects: if it fails, calls setstate(eofbit).


If you have any further questions/comments please contact me directly at:

khurram@metrowerks.com

Thank you
Khurram Qureshi

Metrowerks, Corp.








In article <martinf-2307951632260001@slip204x.slip.colostate.edu>,
martinf@lamar.colostate.edu (Martin Ferguson) wrote:

> Greetings all,
>
> I have just been informed that get(char&) is not supposed to set eofbit
> when it reads the end of file.  Only the get(char* buf, int limit, char
> delim =
> '\n') operation should set the eofbit.  Is this true, and, if so, what is
> the rationale for this?
>
> This just caused me a headache because get(char&) DOES set eofbit on an
> IBM/AIX/xlC, on SUN/Solaris/GNU g++, and on a Mac/Symantec C++ 7.  But NOT
> on a Mac/Metrowerks CW6, which, of course, is where I really want the
> program to work.
>
> Thanks,   Martin

Greetings all,

I have just been informed that get(char&) is not supposed to set eofbit
when it reads the end of file.  Only the get(char* buf, int limit, char
delim =
'\n') operation should set the eofbit.  Is this true, and, if so, what is
the rationale for this?

This just caused me a headache because get(char&) DOES set eofbit on an
IBM/AIX/xlC, on SUN/Solaris/GNU g++, and on a Mac/Symantec C++ 7.  But NOT
on a Mac/Metrowerks CW6, which, of course, is where I really want the
program to work.

Thanks,   Martin