[Bug libstdc++/26181] New: istream::operator>>(streambuf*) fails to set eofbit

[Bug libstdc++/26181] New: istream::operator>>(streambuf*) fails to set eofbit

[sebor at roguewave dot com]

The program below is expected to run successfully to completion.

The extractor must set eofbit even though doing so is not explicitly specified
in 27.6.1.2.3, p12, since it is specified by the blanket statement in 27.6.1.1,
p3.

$ cat t.cpp && g++ -dumpversion && g++ t.cpp -static && ./a.out
#include <cassert>
#include <fstream>
#include <sstream>

int main ()
{
    std::ifstream ifs (__FILE__);
    std::stringstream ss;

    assert (0 == ifs.tellg ());

    ifs >> ss.rdbuf ();

    assert (ifs.eofbit == ifs.rdstate ());
    assert (-1 == ifs.tellg ());

    ifs.clear ();

    assert (0 < ifs.tellg ());

    ifs.get ();

    assert (-1 == ifs.tellg ());
}
4.0.2
Assertion failed: ifs.eofbit == ifs.rdstate (), file t.cpp, line 14
Abort (core dumped)

This first came up here:
http://forum.sun.com/thread.jspa?threadID=28814


-- 
           Summary: istream::operator>>(streambuf*) fails to set eofbit
           Product: gcc
           Version: 4.0.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: sebor at roguewave dot com
 GCC build triplet: all
  GCC host triplet: all
GCC target triplet: all


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-02-09 00:07 -------
It also fails with "4.1.0 20051026".


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pcarlini at suse dot de]

------- Comment #2 from pcarlini at suse dot de  2006-02-09 00:10 -------
Confirmed, thanks a lot!


-- 

pcarlini at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-02-09 00:10:21
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pcarlini at suse dot de]

------- Comment #3 from pcarlini at suse dot de  2006-02-09 11:53 -------
Working on it.


-- 

pcarlini at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pcarlini at suse dot de
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pcarlini at suse dot de]

------- Comment #4 from pcarlini at suse dot de  2006-02-09 17:24 -------
Martin, I have a patch in the making which is supposed to fix the problem.

However, with that patch applied, your testcase still fails the third
assertion, that is 'assert (-1 == ifs.tellg ());' at line 15, because tellg
returns the length of the file, not -1. By the way, Dinkum as bundled together
with ICC does the same.

I will look into that, as soon as I finish the first part of the work, but I
wonder if you could possibly also investigate a bit? Maybe we are diverging in
the implementation (not implementation) of some DRs? Thanks.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[sebor at roguewave dot com]

------- Comment #5 from sebor at roguewave dot com  2006-02-09 17:29 -------
Here's my analysis copied from the post on the Sun C++ Forum.

According to 27.6.1.3, p36a, tellg() behaves as an unformatted input function
(as described in 27.6.1.3, p1).

>From 27.6.1.3, p1: Each unformatted input function begins execution by
constructing an object of class sentry with the default argument noskipws
(second) argument true. If the sentry object returns true, when converted to a
value of type bool, the function endeavors to obtain the requested input.

Since eofbit is set prior to calling tellg() the function must not "endeavor to
obtain the requested input." Hence it must fail.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pcarlini at suse dot de]

------- Comment #6 from pcarlini at suse dot de  2006-02-09 17:35 -------
(In reply to comment #5)
> Here's my analysis copied from the post on the Sun C++ Forum.

[snip]

Ah, now I remember thanks! It's a very old issue, unrelated to the present one,
I remember discussing it briefly with Benjamin so *much* time ago! And I'm
pretty sure we came to the conclusion that indeed your reading of the standard
is the right one, but then didn't take any action, i.e., in practice our tellg
still doesn't have any sentry. If you agree, I'm going to deal with that
separately, however, will file a separate PR myself.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[sebor at roguewave dot com]

------- Comment #7 from sebor at roguewave dot com  2006-02-09 17:43 -------
Sure, whatever works for you :)

FWIW, I find the current behavior in this specific case quite reasonable (i.e.,
it seems reasonable to expect to be able to get the current position even when
the stream has reached the EOF) but that's not what the text says. I also have
a feeling that relaxing the text (and the implementation) to allow this
particular behavior might lead to surprises in some corner cases (I don't have
a test case or anything specific in mind, just a hunch, so it's quite possible
that I'm being overly paranoid).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[paolo at gcc dot gnu dot org]

------- Comment #8 from paolo at gcc dot gnu dot org  2006-02-10 18:29 -------
Subject: Bug 26181

Author: paolo
Date: Fri Feb 10 18:29:04 2006
New Revision: 110841

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=110841
Log:
2006-10-02  Paolo Carlini  <pcarlini@suse.de>

        PR libstdc++/26181
        * include/bits/streambuf.tcc (__copy_streambufs_eof): New, like
        the existing __copy_streambufs but reporting eof in input.
        (__copy_streambufs): Just use the latter.
        * src/streambuf.cc (__copy_streambufs_eof): Adjust specializations
        of __copy_streambufs.
        * include/bits/istream.tcc (operator>>(__streambuf_type*)): Use
        __copy_streambufs_eof instead.
        * include/std/std_streambuf.h: Adjust.
        * src/streambuf-inst.cc: Adjust.
        * config/abi/pre/gnu.ver: Export the new symbols.
        * testsuite/27_io/basic_istream/extractors_other/char/26181.cc: New.
        * testsuite/27_io/basic_istream/extractors_other/wchar_t/26181.cc:
        Likewise.
        * testsuite/27_io/basic_istream/extractors_other/char/1.cc: Adjust.
        * testsuite/27_io/basic_istream/extractors_other/wchar_t/1.cc:
        Likewise.

Added:
   
trunk/libstdc++-v3/testsuite/27_io/basic_istream/extractors_other/char/26181.cc
   
trunk/libstdc++-v3/testsuite/27_io/basic_istream/extractors_other/wchar_t/26181.cc
Modified:
    trunk/libstdc++-v3/ChangeLog
    trunk/libstdc++-v3/config/abi/pre/gnu.ver
    trunk/libstdc++-v3/include/bits/istream.tcc
    trunk/libstdc++-v3/include/bits/streambuf.tcc
    trunk/libstdc++-v3/include/std/std_streambuf.h
    trunk/libstdc++-v3/src/streambuf-inst.cc
    trunk/libstdc++-v3/src/streambuf.cc
    trunk/libstdc++-v3/testsuite/27_io/basic_istream/extractors_other/char/1.cc
   
trunk/libstdc++-v3/testsuite/27_io/basic_istream/extractors_other/wchar_t/1.cc


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181

[Bug libstdc++/26181] istream::operator>>(streambuf*) fails to set eofbit

[pcarlini at suse dot de]

------- Comment #9 from pcarlini at suse dot de  2006-02-10 18:30 -------
Fixed for 4.2.0.


-- 

pcarlini at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26181