From: Warren Young
Date: June 10 2005 9:32pm
Subject: Re: Link errors with resetdb example program
List-Archive: http://lists.mysql.com/plusplus/4507
Message-Id: <42AA06D7.50606@etr-usa.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Dale Smith wrote:
> As you suggested, I commented out everything in resetdb.cpp except for the 
> connection object and tried to compile. 

So, same result, yes?

Can you now try linking a basic "hello, world" type program against 
libmysql++?  If you have to include something in the program to make 
sure it uses the library, use something peripheral within the library, 
like string2stream, or some of the primitive C type stuff in convert.h. 
  You might find the #include graphs in the reference manual helpful 
here: choose something that doesn't eventually depend on query.h, or 
even row.h, connection.h, or coldata.h.

> 000000000008f990 t std::basic_iostream<char, std::char_traits<char> 
> 
>>::~basic_iostream [in-charge deleting]()

That seems to be the method the linker claims it cannot find...

> what is this supposed to show? 

That the method in fact exists within the library.  It is possible that 
whoever built your C++ standard library made it such that it doesn't 
have specializations for iostreams on chars.  The iostream types are 
templates, you know.

> I can't say for sure that the Query object is not implicated, as the above 
> shows the library itself generates link errors.

I don't see that.  What are you referring to, exactly?

Would you be willing to check out MySQL++ v2.0 pre-releases from our 
Subversion repository?  If the changes to Query will happen, it will 
happen during the v2.0 development.  It is not yet certain that it will 
happen, because all known problems presently have reasonable 
workarounds.  So far, the best reason I have to change it is that Scott 
Meyers said so.  Not the best reason...