It isn't the extra-nonce subscription causing issues maybe is it? I do have "no-extranonce-subscribe" : true, in all my profiles as well as globally.
I don't see how that would not allow you to either a) manually switch to another alive pool b) crash out the miner in 60 seconds c) force your starting fanspeeds to 99% after a crash out or d) change anything from the previous build from 06.16.2014
This is kind of shitty now if the entire github repo is polluted with this config parsing bullshit if it broke the miner for so many different people.
Extranonce subscribe used to cause cause sgminer to exit because of bad communication with the pool.... I wrote about this previously (and others have as well). Not saying it's your problem for sure, but I would definitely add "no-extranonce-subscribe" : true to each profile as well as the global config to confirm that doesn't help.
And hey, I am glad the config system is changing. The old one was crap. Give it a few days to stabilize if you're having issues before calling it bad - and keep in mind, it's working great for me on Linux.
Anybody that can post a proven, working config file, with a proven, working logfile of successful algo-flips to all algos (x11/x13/keccak/scrypt/nscrypt) I'd be incredibly interested.
EDIT: And I want to see in the logfile that it can successfully switch from scrypt nf10 to nscrypt nf11 in the same mining instance, without an Error -4 GPU disabled error occurring)
I can post my full config and a log of this tomorrow when I have some time, although I don't see how it will help anything.