boostorg
Teardown javadocs need improvement
The docs for teardown and async_teardown need to make it clear that they are customization points and should not be called directly. See: https://www.boost.org/doc/libs/1_79_0/libs/beast/doc/html/beast/using_websocket/teardown.html
I'm replicating the discussion here for reference.
Question:
Hello gents, I am trying to teardown a websocket stream on the server side, but the compiler complains that no matching function can be found. What am I missing?
websocket::stream<tcp::socket> ws_;
beast::async_teardown(beast::role_type::server, ws_.next_layer(),
[id=id_, on_close=std::move(on_close)](beast::error_code const& ec)
{
on_close(id, ec);
});
Answer:
That function is not meant to be called directly It is a customization point Just call
websocket::stream::async_closeThe docs are terrible But it is better explained here
I see, thanks. Yeah, I read both pages, but came away thinking that I could still call it as long as I provided the underlying socket properly. The thing that lead me down this route is trying to deal with a specific case. Server gets a message and tries to send it over to its final destination - it manages to write the contents using async_write at which point the stream gets closed with async_close - what happens to the written bytes? Are they guaranteed to be sent or will close prevent that? Teardown appeared to do the right thing, although the wording is still a little ambiguous: This operation cleanly discards bytes remaining at receiving endpoints and also closes the underlying TCP/IP connection. If close properly sends the bytes over I am all set, but if not - is there a way for me to guarantee the send?
if you wait for async_write to complete before calling async_close, then the message is guaranteed to be written The comment refers to data received before calling async_close which you did not yet acquire with async_read
