From: Warren Young
Date: April 5 2006 3:03am
Subject: Re: Escaping a string
List-Archive: http://lists.mysql.com/plusplus/5617
Message-Id: <44333384.3050908@etr-usa.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Samuel Borgman wrote:
> 
> I though of a char * which is
> obviously taken care of(by manip.cpp:99-114 I'd imagine).

I looked into this, and it's going to be a mess to fix.

The reason the function you're referring to uses such a tortured 
declaration ("const char* const& in" ... yuck!) is because it is 
specializing a template which expects a const reference to something. 
So, this is trying to talk about a const reference to a const char*. 
Unfortunately, the compiler isn't playing along -- it doesn't consider a 
const char* to be close enough for it to call that specialization. 
Instead, it calls the non-specialized version, which doesn't do quoting. 
  This may be one of those things that once worked, but which was 
"fixed" in a newer compiler version.

If the templates in question are changed to accept any type (i.e. just a 
plain T instead of a const T&), and "const&" is removed from all of the 
char* specializations, string literals get quoted as you'd expect. 
Presumably this will fix Bjoern's problem, as it's probably related to 
the one I illustrated in my program above.  But, this simple change a) 
breaks the ABI; and b) breaks other things, like quoting std::string.

I've added a Wishlist item to the v3.0 section.  In the meantime, I 
suggest wrapping all problem strings in std::string.  It'll create some 
extra temporaries, but what with all the delays imposed by SQL 
databases, I'll bet it won't change your benchmark results meaningfully.