Luca Schlecker

No worries @philhzss, I understood your point. Nice, glad to hear that you got it working, even though you had to deal with something new. Crow currently manages its own...

I would have thought about tying the `crow::request` object to the connection. The handler function would have to use that `crow::request` object it was given and pass it on to...

But I guess this approach isn't all too far from just making every callback async, which would make it even easier to use as one can just block in the...

Just to throw it in there: What about [`std::promise`](https://en.cppreference.com/w/cpp/thread/promise)? One thread would be needed to check on the futures though. I think I'd prefer a mechanism similar to the `crow::finish`...

Well, how about returning a specialized value like `crow::response::async` analogous to `std::nullopt`? If `crow::finish` is then called on a request object where the handler didn't return that special value it...

This wouldn't be too far from the solution @dranikpg and I worked out. I haven't yet looked at any code, including that of `crow::response`.

I will take a look at it, it will take a bit though. Are you still able to use Crow even though that variable is wrong?

I currently cannot confirm this behavior. Could you provide additional information? I tried replicating with the following steps provided on Crow's `Getting Started` page: [Installation instructions](https://crowcpp.org/master/getting_started/setup/linux/#using-cmake) [Consumption instructions](https://crowcpp.org/master/getting_started/setup/linux/#using-cmake_1)

> I'm a little confused as to how the source files are able to use `CHECK` without actually including the Catch2 header.. > > Another thing that interested me is...

As the author of the #457 and the follow-up #482 PR, I'll just throw in my opinion too. I've generally disliked using Boost due to its sheer size. Asio is...