Hi, > While this is expedient, adding dco dependence to parts of code which should > be independent of such inner details look wrong style to me. Finding a way to > keep struct context a const here will lead to a better logic, I think.
We have the same approach in multi_print_status(). Not saying that "that's why this is right", but we should at least be consistent. > Can't we have a coarse timer that keeps the stats updated? It doesn't have to > be frequent as this is just for information. I don't think that "purity" would justify the complexity of adding an additional timer for fetching dco-specific stats. >> + status_printf(so, "TCP/UDP read bytes," counter_format, >> c->c2.link_read_bytes + c->c2.dco_read_bytes); >> + status_printf(so, "TCP/UDP write bytes," counter_format, >> c->c2.link_write_bytes + c->c2.dco_write_bytes); > > Same here -- have a place to read the total count from independent of dco. But what is wrong in fetching dco stats when we need them? What is the advantage of timer? -Lev _______________________________________________ Openvpn-devel mailing list Openvpn-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openvpn-devel
