On Fri, Jul 20, 2018 at 4:01 PM, Ruediger Pluem <[email protected]> wrote: > > You describe the inconsistency in case c->data_in_input_filters is 1, > I did for the case that c->data_in_input_filters is 0, but yes we could have > inconsistencies in both cases > and the question is if they are always fine. So far it looks like. > But probably the flags should just go in order to avoid these cases.
Done in r1836364. Tested/debugged with pipelining, both with event and worker, all seems to work as expected. Regards, Yann.
