Anarthal (Rubén Pérez)

Making the destructor invoke a sync method (that can actually fail) can have even worse consequences: ``` awaitable do_query(tcp_ssl_connection& conn) { results r; auto stmt = co_await conn.async_query("SELECT name FROM...

Sure, in a sync world it's the way to go. > This is besides the point, do you prefer accidental blowup/DOS of production db (global stmt limit) or reduced performance...

That's unfortunately C++20-coroutine-specific, and from what I've read, it looks specific to each coroutine library. This library targets Asio's universal async model. Async functions are not coroutine or callback based...

Raised by #173

I never solved it. MySQL needs hashing functions regardless of whether you're using TLS or not, and reimplementing sha256 wasn't something I wanted. I may change my mind at some...

I'd recommend that you asked in Boost's mailing list first if you haven't before. You can act in consequence depending on the answers you get. If you move forward with...

I think it can be useful, either as a standalone library or as part of asio.

There is no one-liner to determine this, but it's doable. MySQL is a little bit clunky with errors (there are a ton of them and not very well documented). I've...

So something I've seen is that it's difficult to escape from `generic_response` in the general case. It may be my lack of knowledge about Boost.Redis. For example, there is a...

> A way of iterating over the generic_response individual responses Can't you just use `basic_node::depth` to distinguish responses? Every time you encounter a node with `depth == 0`, that's a...