There might be a chance that the device remembers the last mixer setting on startup. So you what you can try is setting the mixer on a windows machine (if you have one available) and see if it's settings are retained on power cycles.
I don't think it does as I generally have monitoring turned off in windows and have been flitting between that and linux throughout my experements. I'll double-check when I'm back at home. I'll also mess with the standalone mode setting as i think one of those does remember the last configuration, so there's a vague chance that it will also mess with the defaults in non-standalone mode.
The other option is to wait until we finish the mixer control. I know that the current code already contains most of the infrastructure to support this, but it's not finished. The difficulty (and opportunity) is that the interface is the same for all bebob devices, requiring it to be very flexible and hence pretty complicated. Daniel is the expert on this, maybe he can shed a light here. Maybe it's not very complicated to have a temporary tool that kills all zero-latency monitoring paths.
Exciting news about the freebob mixer stuff, I hadn't realised this was allready well under way. Will keep my eyes peeled. Additional good news, bad news - Phantom power is not switchable from the front panel unless you are in standalone mode. Once the unit has gone into this mode you can't boot jack until you power cycle the saffire, which switches the phantom power off. Pretty bad news as I imagine phantom power switching is a focusrite not a bebob feature so freebob is unlikely to support it (correct me if i'm wrong) Good news is that the focusrite tech support people were quite understanding and agreed to forward my concerns to their dev team to see if they could get some sort of solution worked out. No promises of course, but i was pleasantly surprised to see that they appeared open to the idea, and it was nice to encounter tech support that not only reads emails but responds to them intelligently. Had a brief flutter with smaller latencies last night, seemed to be pretty stable in realtime mode. The only xrun i got was when i was booting SuperCollider's server, however I *really* knew about it, very loud, fairly long noise burst. The saffire has an assignable volume control which isn't assigned to anything by default so it came out as full scale nastiness. Thankfully i have the sensitivity of my monitors tuned down as far as they go or i could have been collecting my tweeters from the opposite side of the room. Serves me right really, it was a bit idiotic to have them connected when testing something like this, but just thought I'd warn everyone else. My girlfriend heard the burst and thought I'd electricuted myself.. :) Front panel headphone controls work so you can safely test via these without ear damage. Anyway, i digress - I didn't tax my computer that much and only went down to 128 samples/period, but i'm pretty optomistic on the latency front. Still not tried any of the digital gubbins yet - that's my next stop..
