Hi all,
I've deicide to change the distributed transaction sync method in spec -
from three-phase commit to two-phase commit - because of two subjects:
1) As I've been notified, NDB transactions use 2-phase commit criteria
(right?). Considering reusing and unifying is a important objective, reuse
its code could be interesting to ensure a solid and general solution.
2) As I've readed a little more about these 2 methods some days ago, the
general enviroment - and its probable faults - take us to prefer the 2-phase
method robustness.
Any comment?
Thanks
Fabricio Mota
