Bryan W. Headley wrote:
>
> you're referring to MysqlRow::operator[](const char*), right?
Yes. There were also std::string and int overloads.
The problem comes up when you say:
bla = row[0];
Which overload does it call? You can convert 0 to all four types, so
g++ rightfully said "I give up". I therefore removed three of the
overloads (all but size_type, which is the only one the library itself
requires) and added MysqlRow::lookup_by_name() to replace the
functionality provided by the char* and string overloads.
By the way, if someone wants lookup_by_name() overloaded for std::string
as well as const char*, let me know. I only stopped with const char*
because that was the minimum change that would allow the examples to
compile.
I should point out that this problem is most acute with 0 as the
argument to operator[]. When you use a different number, you rule out
the const char* overload. When you use a variable, you rule out all but
one, if the variable is of the right type. Therefore, it's quite
possible for the old code to compile on recent g++'s without this
change, depending on how you actually used the library. It happens that
the SQLSS code uses bare constants, including 0, so all custom SQL macro
users hit this problem. If you never built the examples and you don't
use the SQLSS stuff, you might never have seen this problem.
> - NAME##_cus_value_list (const NAME *o, cchar *d, Manip m ,vector<bool>)
> + NAME##_cus_value_list (const NAME *o, cchar *d, Manip m
> ,vector<bool> *i)
That's indeed the change I applied to my working copy. It will be in
version .14, which may be out soon. I'm working on other things which
may also go into that release, and I don't know how long it will take.
