On Mon, 12.01.09 17:15, Colin Guthrie (gmane at colin.guthr.ie) wrote: > > Hmm, PA_SINK_NETWORK was introduced for this very purpose. It is a > > flag you can set if your sink is already wrapping some kind of network > > device and hence it shouldn't be wrapped a second time. It doesn't > > really matter if the protocol in question is our native protocol > > (i.e. as in "tunneling") or is RAOP, or UPnP or whatever. We generally > > should say that network sinks should always connect "as directly" as > > possible. > > > I feel entirely unoriginal :p > > Just out of interest, where do you see flags vs. props going? Is it > worth trying to consolidate them a bit or is having both systems (which > arguably could do the same thing in many cases) still sensible for some > reason? We need to keep compat for starters. However, even if that wasn't a problem i'd probably keep both props and the flags field. Flags is more for low-level control of a few things that are relevant to the core itself. OTOH the props stuff is for a higher-level type of policy which is not enforced by the core itself but by add-on policy modules. The line between these kinds of policy is a blurred of course, but I think clear enough to be a guide for future editions. Lennart -- Lennart Poettering Red Hat, Inc. lennart [at] poettering [dot] net ICQ# 11060553 http://0pointer.net/lennart/ GnuPG 0x1A015CC4
