[Bug libstdc++/49384] New: istringstream::tellg at the end of string gives a wrong result

[Bug libstdc++/49384] New: istringstream::tellg at the end of string gives a wrong result

[pogonyshev at gmx dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49384

           Summary: istringstream::tellg at the end of string gives a
                    wrong result
           Product: gcc
           Version: 4.6.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: pogonyshev@gmx.net


According to the specification:

    Returns: After constructing a sentry object, if fail() != false, returns
pos_type(-1) to indicate failure. Otherwise, returns rdbuf()->pubseekoff(0,
cur, in).

Tested with geordi:

    { istringstream x ("lol"); string y; x >> y; cout << y, x.fail (); cout,
x.tellg (); cout, x.rdbuf ()->pubseekoff (0, ios::cur, ios::in); }

Output:
    lol, false, -1, 3

Note that fail() returns 'false', yet tellg() gives -1.  Also, I tested the
standard-prescribed return value above: it is 3 as expected, suggesting that
tellg() tests eof() rather than fail() to decide if return -1 or not.

This used to work as standardized before (I believe with 4.6.0 too, though not
sure).  I noticed after my program stopped working correctly without any
changes, just compiler upgrade.

[Bug libstdc++/49384] istringstream::tellg at the end of string gives a wrong result

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49384

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID

--- Comment #1 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-06-12 18:54:52 UTC ---
The new behavior is intended, see this exchange for further details:

  http://gcc.gnu.org/ml/libstdc++/2011-04/msg00026.html

[Bug libstdc++/49384] istringstream::tellg at the end of string gives a wrong result

[pogonyshev at gmx dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49384

--- Comment #2 from Paul Pogonyshev <pogonyshev at gmx dot net> 2011-06-13 20:09:49 UTC ---
So, changing in a way incompatible to what the standard says is intended?  Or
am I (and pre-4.6 libstdc++) misreading the standard?

[Bug libstdc++/49384] istringstream::tellg at the end of string gives a wrong result

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49384

--- Comment #3 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-06-13 21:11:57 UTC ---
The published C++ Standard has DEFECTS, as any other Standard. With time,
defects are analyzed, fixes found (which then become part of the Standard
actually in force) and then implemented by the implementors, like GCC.