[Bug libstdc++/61101] New: snprintf not part of std

[Bug libstdc++/61101] New: snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

            Bug ID: 61101
           Summary: snprintf not part of std
           Product: gcc
           Version: 4.8.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: maxime.boissonneault at calculquebec dot ca

Created attachment 32755
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32755&action=edit
The output with -save-temps

- the exact version of GCC; 
  GCC 4.8.2
- the system type; 
  CentOS 6.5
- the options given when GCC was configured/built;
  ../${SRCDIR}/configure CFLAGS="$FLAGS" CXXFLAGS="$FLAGS" \
  LDFLAGS="-L$PREFIX/lib" --prefix=$PREFIX \
  --with-arch=native --with-pkgversion=$VERSION-colosse \
  --enable-languages=c,c++,java,fortran,go --enable-shared \
  --enable-static --enable-threads=posix --enable-multilib \
  --enable-bootstrap --with-gmp=$PREFIX --with-mpfr=$PREFIX --with-mpc=$PREFIX
\
  --with-system-zlib

- the complete command line that triggers the bug;
   g++ test.cpp -std=c++11

- the compiler output (error messages, warnings, etc.); and
  the preprocessed file (*.i*) that triggers the bug, generated by adding
-save-temps to the complete compilation command, or, in the case of a bug
report for the GNAT front end, a complete set of source files (see below).

test.cpp: In function ‘int main()’:
test.cpp:5:2: erreur: ‘snprintf’ is not a member of ‘std’
  std::snprintf(test, 10, "%f", 2.);
  ^
test.cpp:5:2: note: suggested alternative:
In file included from
/software6/compilers/gcc/4.8.2/include/c++/4.8.2/cstdio:42:0,
                 from test.cpp:1:
/usr/include/stdio.h:385:12: note:   ‘snprintf’
 extern int snprintf (char *__restrict __s, size_t __maxlen,
>From gcc-bugs-return-450888-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Wed May 07 19:01:05 2014
Return-Path: <gcc-bugs-return-450888-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 10030 invoked by alias); 7 May 2014 19:01:05 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 9973 invoked by uid 48); 7 May 2014 19:01:00 -0000
From: "olegendo at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/61026] sh-rtems4.11 build of 4.9.0 fails on FreeBSD 10 c++ (clang).
Date: Wed, 07 May 2014 19:01:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: olegendo at gcc dot gnu.org
X-Bugzilla-Status: RESOLVED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: olegendo at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-61026-4-n4yYdt8JA2@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-61026-4@http.gcc.gnu.org/bugzilla/>
References: <bug-61026-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-05/txt/msg00580.txt.bz2
Content-length: 367

http://gcc.gnu.org/bugzilla/show_bug.cgi?ida026

--- Comment #10 from Oleg Endo <olegendo at gcc dot gnu.org> ---
Sorry for the trouble, but Joseph has a point here ...
http://gcc.gnu.org/ml/gcc-patches/2014-05/msg00414.html

Could you guys please try moving the problematic header includes in sh.c below
the #include "config.h" and report your findings?  Thanks.

[Bug libstdc++/61101] snprintf not part of std

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> ---
What was the value of $FLAGS when GCC was built?

Something is preventing the library from using C99 functions.

What is the output of running this?

grep USE_C99
/software6/compilers/gcc/4.8.2/include/c++/4.8.2/x86_64-unknown-linux-gnu/bits/c++config.h

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #2 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
[mboisson@colosse3 ~]$ grep USE_C99
/software6/compilers/gcc/4.8.2/include/c++/4.8.2/x86_64-unknown-linux-gnu/bits/c++config.h
/* #undef _GLIBCXX_USE_C99 */
/* #undef _GLIBCXX_USE_C99_COMPLEX */
/* #undef _GLIBCXX_USE_C99_COMPLEX_TR1 */
#define _GLIBCXX_USE_C99_CTYPE_TR1 1
/* #undef _GLIBCXX_USE_C99_FENV_TR1 */
#define _GLIBCXX_USE_C99_INTTYPES_TR1 1
#define _GLIBCXX_USE_C99_INTTYPES_WCHAR_T_TR1 1
#define _GLIBCXX_USE_C99_MATH 1
#define _GLIBCXX_USE_C99_MATH_TR1 1
#define _GLIBCXX_USE_C99_STDINT_TR1 1

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #3 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Sorry, I missed the first question. The value of $FLAGS was
FLAGS="-O2 -march=native -pipe -Wall"

[Bug libstdc++/61101] snprintf not part of std

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Maxime Boissonneault from comment #2)
> /* #undef _GLIBCXX_USE_C99 */

Something is wrong with your GCC build.

I've just installed it on CentOS 6.5 and it works fine.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #5 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Any pointer on what could be wrong ?

Maybe a clue is that we compiled GCC not with the system GCC, but with another
GCC ?

Basically, we started with the system GCC. We compiled GCC, GMP, MPFR, MPC_V,
with that system GCC. Then, we uninstalled the system GCC and its libstdc++ and
recompiled those with the GCC we had just compiled.

>From then on, we just used that GCC to compile the next one (we started with
4.8.0, then 4.8.2).

Maxime

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #6 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Here is our complete installation document : 
https://docs.google.com/a/calculquebec.ca/document/d/1hcddCXGnm6OgTwRxRDU2akb-AxKx2aursppLp2JPwWo/edit

It went :
- System GCC compiled GCC 4.8.1
system GCC was removed.

- GCC 4.8.1 compiled GCC 4.8.2
- GCC 4.8.2 compiled GCC 4.6.4

Each time we bootstrapped, so that in the end, I believe GCC 4.8.2 was compiled
with 4.8.2, etc.

[Bug libstdc++/61101] snprintf not part of std

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #7 from Paolo Carlini <paolo.carlini at oracle dot com> ---
I would suggest having a look to the config.log in the libstdc++-v3 build
directory and see which and how a C99 test is failing.

By the way (Jon) in mine I see a warning:

conftest.cpp: In function 'int main()':
conftest.cpp:54:23: warning: deprecated conversion from string constant to
'char*' [-Wwrite-strings]
 snprintf("12", 0, "%i");
                       ^
configure:16942: $? = 0
configure:16953: result: yes

which we should probably avoid.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #8 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Created attachment 32758
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32758&action=edit
config.log, configure.log, make.log

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #9 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
I attached the config.log, configure.log and make.log. The only one containing
C99 is the make.log. 

The only test that says no is

checking for fully enabled ISO C99 support... no

The rest are yesses :
checking for ISO C99 support in <math.h>... yes
checking for ISO C99 support in <stdio.h>... yes
checking for ISO C99 support in <stdlib.h>... yes
checking for ISO C99 support in <wchar.h>... yes
checking for fully enabled ISO C99 support... no
checking for ISO C99 support to TR1 in <ctype.h>... yes
checking for ISO C99 support to TR1 in <stdint.h>... yes
checking for ISO C99 support to TR1 in <math.h>... yes
checking for ISO C99 support to TR1 in <inttypes.h>... yes
checking for wchar_t ISO C99 support to TR1 in <inttypes.h>... yes

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #10 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Created attachment 32759
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32759&action=edit
stage1-libstdc++-v3-config.log

Here is there libstdc++-v3 config.log file.

[Bug libstdc++/61101] snprintf not part of std

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #11 from Paolo Carlini <paolo.carlini at oracle dot com> ---
But that is what matters, because the stdio functions in the header are
protected by the global C99 macro, not by something more fine grained. In any
case, on any normal Linux system the final outcome must be yes, thus something
is going wrong.

[Bug libstdc++/61101] snprintf not part of std

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #12 from Paolo Carlini <paolo.carlini at oracle dot com> ---
In your log there are a lot of errors: "fatal error: bits/c++config.h: No such
file or directory" which are definitely unexpected and bad.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #13 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
I see that. 

Wasn't the GCC 4.8.1 compiler supposed to already know to look in the folder
/software6/compilers/gcc/4.8.1/include/c++/4.8.1/x86_64-unknown-linux-gnu/
to find bits/c++config.h ?

Do I need to manually add that to the CPATH ?

[Bug libstdc++/61101] snprintf not part of std

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #14 from Paolo Carlini <paolo.carlini at oracle dot com> ---
You can experiment at will, but remember that on any sane Linux system nothing
similar is necessary, a tarball or a checkout can be built as is and all the
C99 facilities are normally enabled. Something I would also recommend for your
experiments, tentatively disable bootstrap, thus build with --disable-bootstrap
and also enable only the languages you need, thus c++, everything will be
quicker and simpler, fewer files will be around.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #15 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
Thanks for the advices. I will try to sort this out tomorrow. The reason we
enable all languages and why we build from source is to control all
dependencies and are able to provide multiple versions of the compilers. I
manage a supercomputer with hundreds of users with various needs.

The reason we bootstrap is to remove as many dependencies as possible on the
OS-provided packages. We've been bitten in the past with updates of packages
(even on CentOS6) that broke user's codes.

Obviously, we can screw up too. 

Thanks again.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #16 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
On a related not, is there a way to control default paths that GCC will always
look into for libraries and/or headers ? We install things in non-standard
locations (so that we can have multiple version and not depend on the OS) and
that would be handy.

[Bug libstdc++/61101] snprintf not part of std

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #17 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Maxime Boissonneault from comment #5)
> Basically, we started with the system GCC. We compiled GCC, GMP, MPFR,
> MPC_V, with that system GCC. Then, we uninstalled the system GCC and its
> libstdc++ and recompiled those with the GCC we had just compiled.

This sounds like a big waste of time.

(In reply to Maxime Boissonneault from comment #6)
> Here is our complete installation document : 
> https://docs.google.com/a/calculquebec.ca/document/d/
> 1hcddCXGnm6OgTwRxRDU2akb-AxKx2aursppLp2JPwWo/edit
> 
> It went :
> - System GCC compiled GCC 4.8.1
> system GCC was removed.

This step might have broken your system.

> - GCC 4.8.1 compiled GCC 4.8.2
> - GCC 4.8.2 compiled GCC 4.6.4
> 
> Each time we bootstrapped, so that in the end, I believe GCC 4.8.2 was
> compiled with 4.8.2, etc.

(In reply to Maxime Boissonneault from comment #15)
> The reason we bootstrap is to remove as many dependencies as possible on the
> OS-provided packages. We've been bitten in the past with updates of packages
> (even on CentOS6) that broke user's codes.

GCC should have almost no dependencies on system libraries, just libc.so
(although I note you've used --with-system-zlib, so that adds one dependency)

Installing GMP, MPFR and MPC instead of linking statically also adds avoidable
dependencies.

[Bug libstdc++/61101] snprintf not part of std

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #18 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Maxime Boissonneault from comment #13)
> I see that. 
> 
> Wasn't the GCC 4.8.1 compiler supposed to already know to look in the folder
> /software6/compilers/gcc/4.8.1/include/c++/4.8.1/x86_64-unknown-linux-gnu/
> to find bits/c++config.h ?
> 
> Do I need to manually add that to the CPATH ?

No. It should be found, so something is wrong.

[Bug libstdc++/61101] snprintf not part of std

[maxime.boissonneault at calculquebec dot ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> changed:

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

--- Comment #20 from Maxime Boissonneault <maxime.boissonneault at calculquebec dot ca> ---
I believe I found the problem. It was an environment variable
CPLUS_INCLUDE_PATH=/software6/compilers/gcc/4.8.1/include/c++/4.8.1

that was set before compiling. Removing this variable seems to make xgcc able
to find c++config.h and all C99 support is enabled.

I'm going to mark this one as resolved invalid.

Thanks again for the insights.

Maxime

[Bug libstdc++/61101] snprintf not part of std

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61101

--- Comment #21 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Thanks for debugging and resolving it.

(In reply to Maxime Boissonneault from comment #0)
> - the options given when GCC was configured/built;
>   ../${SRCDIR}/configure CFLAGS="$FLAGS" CXXFLAGS="$FLAGS" \
>   LDFLAGS="-L$PREFIX/lib"

N.B. I don't think this LDFLAGS should be needed either.