[MySQL-plusplus] MySQL++ not working with libstdc++6

I believe this is not correct. Assigning something with a basic_ios
inside never works.

Try:

mysqlpp::Query q(connection.query("SELECT * from Meter"));

This is exactly equivalent.

The error seems to indicate a race condition in the initialization of
the std::ios_base base class. I've seen someone report a similar
error in that libstdc++ code before, but I think it's correct (I can't
see a problem anyway) so it could be user error, but I'm not sure.

But the same code has been working without problems before the libstdc++6
was updated. But any suggestions how to fix it?? Or avoid it?? Should I use
the other constructor for a Query?? It's pretty obvious I have to create a
query to get the data from the MySQL server.

-Ina

2011/8/19 Jonathan Wakely <[email protected]>

I reinstalled mysql++ and g++ and libstdc++, but no luck with those. Then I
made a simple main program from the example simple1.cpp

int main()
{

mysqlpp::Connection conn(false);
if (conn.connect(DBNAME, DBADDRESS, DBUSER, DBPWD, DBPORT)) {
// Retrieve a subset of the sample stock table set up by resetdb
// and display it.
mysqlpp::Query query = conn.query("SELECT * from Meter");
} else {
cerr << "DB connection failed: " << conn.error() << endl;
return 1;
}

return 1;
}


And valgrind says this:

==24122== Use of uninitialised value of size 8
==24122== at 0x54FE181: std::locale::operator=(std::locale const&) (in
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x54FCC6E: std::ios_base::_M_init() (in
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x55117C8: std::basic_ios<char, std::char_traits<char>
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x526D163: mysqlpp::Query::Query(mysqlpp::Connection*, bool,
char const*) (ostream:83)
==24122== by 0x525CA4F: mysqlpp::Connection::query(char const*)
(connection.cpp:264)
==24122== by 0x44152F: main (main.cc:29)
==24122== Uninitialised value was created by a stack allocation
==24122== at 0x4414AF: main (main.cc:22)
==24122==
==24122==
==24122== ---- Attach to debugger ? --- [Return/N/n/Y/y/C/c] ---- c
==24122==
==24122== Process terminating with default action of signal 11 (SIGSEGV)
==24122== Bad permissions for mapped region at address 0x441699
==24122== at 0x54FE181: std::locale::operator=(std::locale const&) (in
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x54FCC6E: std::ios_base::_M_init() (in
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x55117C8: std::basic_ios<char, std::char_traits<char>
/usr/lib/libstdc++.so.6.0.13)
==24122== by 0x526D163: mysqlpp::Query::Query(mysqlpp::Connection*, bool,
char const*) (ostream:83)
==24122== by 0x525CA4F: mysqlpp::Connection::query(char const*)
(connection.cpp:264)
==24122== by 0x44152F: main (main.cc:29)
==24122==
==24122== HEAP SUMMARY:
==24122== in use at exit: 94,177 bytes in 69 blocks
==24122== total heap usage: 119 allocs, 50 frees, 117,895 bytes allocated

I'd say this puts the error to the mysql++ side??

-Ina

2011/8/21 Ina J. <[email protected]>

Ina

Can you provide the full code for your sample and the compile line that you
are using. Please modify the sample to use one of the mysql++ example
tables.

I'll download Ubuntu to see if I can reproduce it.



Adrian

It may be important that this is on Ubuntu 10.4 LTS, and the current
stable source release rather than what is in the Ubuntu package repo.

That could also explain why no one else is complaining about this: I
suspect most of our Ubuntu users aren't still on the LTS branch, and are
either using the packaged MySQL++ (3.0.9 still) or svn.

Ina: have you tried the svn version?

Ina,

I cannot reproduce you issue - can you check the details below and see if I
have missed anything?

I downloaded and installed
[email protected]:~$ uname -a
Linux unbuntu 2.6.32-33-generic #70-Ubuntu SMP Thu Jul 7 21:13:52 UTC 2011
x86_64 GNU/Linux

I tried mysqlpp 3.0.9 from ubuntu packages and its worked fine.

I removed 3.0.9 and installed 3.1.0 from source and it worked fine as well

Out of interest - what does this version code do on you box
[email protected]:~$ g++ -o version version.cc -I/usr/local/include -lmysqlpp
`mysql_config --include --libs`
[email protected]:~$ ./version
[email protected]:~$ cat version.cc
#include <mysql++/mysql++.h>

int main()
{
if(mysqlpp::get_library_version()!=MYSQLPP_HEADER_VERSION)
{
std::cout << "Library/header mismatch\n";
}
return 0;
}


Detailed results of my testings

[email protected]:~$ g++ --version
g++ (Ubuntu 4.4.3-4ubuntu5) 4.4.3
Copyright (C) 2009 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[email protected]:~$ ls -l /usr/lib64/libstdc++.so.6.0.13
-rw-r--r-- 1 root root 1044112 2010-03-26 18:16
/usr/lib64/libstdc++.so.6.0.13
[email protected]:~$ grep MYSQLPP_HEADER_VERSION /usr/include/mysql++/mysql++.h

#define MYSQLPP_HEADER_VERSION MYSQLPP_VERSION(3, 0, 9)
/// MYSQLPP_HEADER_VERSION constant in order to ensure that your
[email protected]:~$ cat simple.cc
#include <mysql++.h>
#include <iostream>

using namespace std;

int main()
{
std::cout << "Version: " << mysqlpp::get_library_version() << std::endl;
mysqlpp::Connection conn(false);
if (conn.connect("mysql_cpp_data", "localhost", "adrian", "", 3306)) {
// Retrieve a subset of the sample stock table set up by resetdb
// and display it.
mysqlpp::Query query = conn.query("SELECT * from stock");
} else {
cerr << "DB connection failed: " << conn.error() << endl;
return 1;
}

return 1;
}
adri[email protected]:~$ g++ -o simple simple.cc -I/usr/include/mysql++
`mysql_config --libs --include` -lmysqlpp
[email protected]:~$ ./simple
Version: 196617
[email protected]:~$

[email protected]:~$ sudo ldconfig -p | grep mysql
libmysqlpp.so.3 (libc6,x86-64) => /usr/local/lib/libmysqlpp.so.3
libmysqlpp.so (libc6,x86-64) => /usr/local/lib/libmysqlpp.so
libmysqlclient_r.so.16 (libc6,x86-64) => /usr/lib/libmysqlclient_r.so.16
libmysqlclient_r.so (libc6,x86-64) => /usr/lib/libmysqlclient_r.so
libmysqlclient.so.16 (libc6,x86-64) => /usr/lib/libmysqlclient.so.16
libmysqlclient.so (libc6,x86-64) => /usr/lib/libmysqlclient.so
[email protected]:~$ grep MYSQLPP_HEADER_VERSION
/usr/local/include/mysql++/mysql++.h
#define MYSQLPP_HEADER_VERSION MYSQLPP_VERSION(3, 1, 0)
/// MYSQLPP_HEADER_VERSION constant in order to ensure that your
[email protected]:~$ g++ -o simple simple.cc -I/usr/local/include/mysql++
`mysql_config --libs --include` -lmysqlpp
[email protected]:~$ ./simple
Version: 196864
[email protected]:~$

Warren: I haven't tried the SVN version out

Adrian: The examples you provided work just fine so the error has to be in
my code structure.

-Ina

2011/8/22 Adrian Cornish <[email protected]>

Found the problem! It was in my compile options.

My makefile's compile section looked like this:

#Compiler
CC=g++
#Compiler options
CCFLAGS=-std=c++98 -Wall -pedantic -Wextra -g -D_GLIBCXX_DEBUG -pthread

#MySQL libs
MYSQL_INC = -I/usr/include/mysql -I/usr/include/mysql++
MYSQL_LIB = -Bsymbolic-functions -rdynamic -L/usr/lib/mysql++ -lmysqlclient
-lmysqlpp


My makefile had an error in it and it didn't use the CCFLAGS at all at the
beginning of the project. I corrected it during the project and then the
problem occured. When I realized this, I started to cut down the options and
found the 'D_GLIBCXX_DEBUG' will trigger the valgrind. I removed it from my
makefile and everything works as it should now.

Why does this option cause that kind of a problem?? Can't even remember
where I found the option and why did I add it after all.

-Ina

PS. I haven't configured/compiled the MySQL++ library with threads enabled,
but the library isn't used in the threads so can I continue like this or
should I still compile it with the threads enabled flag?? Only the main
program uses the library.

It turns on the libstdc++ debug mode, but you have to use it
consistently for everything, you can't generally mix objects built
with it and objects built without it.

Okay. Thanks for the answer. Those compiler options are used in the whole
project so leaving it out is not a problem. All the objects will be compiled
without it.

-Ina

2011/8/23 Jonathan Wakely <[email protected]>