By the definition of Disconnecting Means.
Well that's a stretch in my opinion. You can't make sense of any of the uses of 'Disconnecting Means' in article 230 if you insist on reading the definition so literally.
Were not talking about the service conductors that enter the cabinet. They don't feed-through the cabinet. I'm discussing the Supply Side Connected inverter output conductors of a interactive system that splice onto the service conductors and feed-through the cabinet to the PV Disconnecting Means. Yes, there is no way to turn off power to these conductors...thats the danger.
Actually, one could argue that they don't feed through the cabinet but rather terminate at the splice. The code is not clear on the proper category for the supply side connected PV conductors, as we know from the dragged-out arguments we've had about it on this forum. The NFPA called them a feeder, so clearly the feeder ends where it hits the service conductors. To be clear, I think this is a semantic argument; ultimately it doesn't matter what words we name them with, what matters is safety. And I remain unconvinced that splicing in an existing main service panel adds any danger that doesn't already exist by having service entrance conductors field installed to a main disconnect inside the enclosure (which is undeniably permitted).
I've seen some service conductors that wind their way through three quarters of a service panel before they land on the main breaker. When the deadfront is off, that's a lot more exposed, unprotected, non-disconnectable conductor than on many of my supply-side taps. And yet, it's not a violation, and it doesn't need to be.
Every panelboard has a warning label to caution about disconnecting power before working on it. Its a real world out there and unqualified persons do open cabinets to work inside. The problem I see is that these supply side connected conductors do more than just land at the Main Service Disconnect. They splice onto those service conductors and feed-through that cabinet without any overcurrent or fault current protection and no disconnecting means other than pulling the meter or cutting the service at the service point. This practice, IMO is against 90.1, practical safeguarding.
To reiterate, nothing you've described about the supply-side connected conductors doesn't also apply to the previously existing service conductors. There is no additional danger that wasn't already there. The service conductors in the enclosure, which also don't have overcurrent or fault protection, are just as dangerous. If someone doesn't understand that those conductors remain energized, then I don't believe any amount of labeling or requiring conductors to be routed any which way is going to keep such a person safe. Such a person shouldn't be working in the panel.
The 705.12(A) Supply Side Connection should be made outside of the cabinet.
I don't think your interpreted requirement addresses the safety issue you're concerned about. An unqualified person could
still think that a 'main disconnect' de-energizes the conductors in the separate PV disconnect. By requiring the splice to be hidden in another enclosure you potentially add to the confusion and difficulty of someone figuring out the installation, qualified or not. For that matter, even a
qualified person could think that a 'service disconnect' de-energizes the whole building, especially if the separate PV disconnect is not also labeled as such. That, after all, is what service disconnecting means are supposed to do. Tangent: this is one of the reasons I think the code should take sides and call supply-side PV conductors service entrance conductors and require a supply side disconnect to be labeled as a 'service disconnect'. That's where I'd invoke 90.1, if it were up to me.
To sum up, I disagree. And I certainly hope that the many AHJs who've permitted me to tap in the main service panel don't change their minds and require much, much costlier work in order to meet an interpretation which does
not, in my opinion, increase anyone's safety one iota.