TCPConnection.waitForDataAsync

[FraMecca]

I have implemented your suggested changes and I am hoping that we can discuss this PR more if needed.

[s-ludwig]

The two lines from above

assert(m_context.readBuffer.length == 0);
m_context.readBuffer.putN(nbytes);``

should be executed right after the tm.stop(), so that the read data is actually committed to the read buffer. The if condition would then ideally be m_context.readBuffer.length > 0 instead of st == IOStatus.ok, because partial data could have been read before the error was raised (not a realistic case, since IOMode.once is used, but handling it this way should be slightly more robust).

[s-ludwig]

To reduce the amount of text to type when using this from the outside, I'd move this outside below the TCPConnection class.

[s-ludwig]

Okay, thanks, that looks good now.

I was thinking whether the runTask is actually such a good idea. Since someone might already have a task running in the background and would just notify it from within the callback, a new task may not always be needed and would just be overhead. For that reason it may be the better solution to just document that no blocking operations should be performed form the callback, since it may be called outside of a task, and to replace the runTask calls by plain callback calls. I'll just quickly add a commit to the PR for that and merge, so that this can still get into the next release.