I cannot agree that the code would allow you to, say, have twelve branch circuit disconnects as the onlt means of disconnecting the building simply because there are two "services."
As far as I'm concerned, there's only one 'service' when there's only one distribution system within the building - np matter how many sources of supply there might be to that system. I will concede that the code language is conflicting on this point.
The conflict is seen in the wording of the section that discussed the number of means of disconnection. It is clear that the authors imagined separate systems - for example, where a 480/277 service served 3-phase loads and a 120/240 service served single-phase loads. I cannot accept that the authors ever imagined both services feeding the same panel.
A similar situation might be where a generator is used- except that the traditional transfer switch was designed to ensure that only one of the sources was ever used at one time.
