Rocket thread exhaustion on timeouts

[Dave S]

On Windows, when using the console (TKInter? IP selection, admin pw, etc), Rocket will run out of threads if you get a bunch (a bunch but not a lot) of timeouts while sending the payload.  This is because the thread that times out will pop up an error window and wait for you to click "OK".

This has been a problem for me for many versions of web2py, currently 

web2py™ Version 2.22.5-stable+timestamp.2022.06.04.18.13.51
Python Python 3.10.7

This seems to be related to log entries like

Rocket.Errors.Thread-21 - ERROR - Tried to send "500 Server Error" to client but received socket error

and my debugging suggests that part of the action is near  read_request_line(), although I haven't finished pushing my probes in here.  Note that clicking "OK" dismisses the pop-up, and makes a thread available until the next timeout.

I admit to setting things up by allowing the response to try to stuff a lot of bits onto the client (I need to follow up on my studies of ts-file splitting), but it still seems to be a Rocket problem.  

This is a small development machine, and I usually don't run without a console ... that would eliminate the pop-ups, I presume, but still may not be graceful handling of timeouts.  I also haven't tried on Linux with or without a console (my Linux projects don't currently have big files to stuff down).  Rocket is also used with py4web AIUI, so timeouts could possibly be an issue there, although the pop-ups might not be part of that environment.

/dps

How much