Thank you for the clarification Jonathan.
On Dec 27, 2:51 pm, Jonathan Lundell <[email protected]> wrote: > By way of explanation... > > My parse_url patch today is strictly refactoring, with no functional > change--another in a series of patches to make URL rewriting more modular. > > The Rocket patch is slightly more than cosmetic, in a development context. > New worker threads, beyond the first default thread pool, did not have > setDaemon set to True. This ordinarily is of no consequence. However, if you > had a request running in one of those threads, and the request got stuck in a > loop (hung), then sending SIGTERM to web2py would fail to terminate that > thread, and you'd have to use SIGKILL (kill -9) to get web2py to quit. > > setDaemon tells the thread logic that the thread in question can be > terminated without waiting for it to stop itself, which is what we want in > this case. > > This is only likely to happen in testing, I imagine, and even then only if > Rocket needed to create additional worker threads, and even then only if the > hung process was running in one of the newer threads. So not a big deal....
