Re: Review of Boost.MySql

From: Ruben Perez (rubenperez038_at_[hidden])
Date: 2022-05-18 10:27:47

On Wed, 18 May 2022 at 11:05, Kostas Savvidis <ksavvidis_at_[hidden]> wrote:
>
> Hi Ruben,
>
>     Some people have already suggested making the tcp_ssl connection as the default. In that spirit, it seems to me that it would be better to
> remove such prefixes as tcp_ssl from tcp_ssl_prepared_statement and tcp_ssl_resultset
> and instead add apropriate prefixes to the non-default connections. Actually, since the only other type is unix socket,
> one only need to add the prefix plaintext or nonssl for the noncrypted connections.

I like how this sounds. I've updated https://github.com/anarthal/mysql/issues/73
to reflect your comments. Note that we may want to add named_pipe_connection
for Windows in the future.

>      Moreover, is there no technical solution that can completely remove the need to deal with these distinct names for resultset and prepared_statement after the appropriate connection is established?

They depend on the Stream type because they have I/O operations. E.g.
for a prepared_statement, you need the Stream type to implement
the prepared_statement::execute function. Internally, these are really
just pointers to the connection object (more or less). One possible
option is making prepared_statement and resultset not be I/O objects,
and move these functions to the connection object. So the following signatures:

prepared_statement::execute(const params<ValueIterator>&)
prepared_statement::close()
resultset::read_one(row&)

Would change to these ones:

connection::execute_statement(const prepared_statement&, const
params<ValueIterator>&)
connection::close_statement(const prepared_statement&)
connection::read_row(resultset&, row&)

What do you think?

Regards,
Ruben.