> Also, it seems that in the deferred lingering case we should probaly
> shorten the socket timeout before calling (and possibly blocking on)
> ap_start_lingering_close()'s hooks/flush, since we likely come from a
> time-up already...

+1

Reply via email to