Waba wrote:
> Here is my patch that enables one to use nullable fields in SQLSS.
So it's only 21 months since you posted this, but it's applied now. :)
Not even close to applied as-is, but I hope I've kept the purpose and
spirit of the patch intact. Some of the differences are because of
changes in the library infrastructure since the v2.0 days when you
created this patch. You took MySQL++ to task for ColData deriving from
std::string, for example, which is no longer true in v3.
Another reason for differences is because I thought I had a better way
to get some things done. The biggest example here is the dancing around
you had to do to make it compile after changing DateTime's ctors to be
templatized for anything that looks like a std::string instead of having
explicit ctors for std::string and ColData. I took the ctor change
idea, which was sound, but made the new templatized ctor explicit. This
avoided the dancing around at the expense of implicit conversions of
string to DateTime. Like many implicit conversion ideas, it sounds good
on paper until you see all the demented ways the compiler has for making
use of this freedom. :)
Anyone reading this who wants to use Null<T> types in SSQLSes: please
try this and report to the list if it doesn't work. This is a very
invasive patch, and I'm not convinced it's completely kosher. You can
check out the svn version of MySQL++ like so:
svn co svn://svn.gna.org/svn/mysqlpp/trunk mysqlp
