There are currently two BIPs floating to enable a "user activated softfork" of Segregated Witness.
BIP148 is meant to activate by flag day on August 1st 2017 whereas enforcing nodes stop accepting blocks that don't signal readiness for segwit.
BIP149 proposes to initiate a second activation process based on BIP8 once the current versionbit (BIP9) activation times out unsuccessfully. From my understanding, BIP8 is the same as BIP9 except that at the end of the activation cycle it goes into
LOCKED_IN instead of timing out.
Isn't it just a semantic difference, whether a subset of all nodes defines the proposal as
LOCKED_IN or the same subset of nodes outright rejects non-signalling blocks until the proposal goes to
What makes BIP149 safer to activate?
I don't see a significant difference or that the former would decrease the chance of a chainsplit significantly except for the longer waiting time until it is activated.