ikelos

Again, I don't see support for only-just published kernels as needing to be rushed into this version rather than waiting for the next one in 3-4 months. It doesn't feel...

So... I'm not sure I'm keen on us just looping forever, and doing all that maths manually isn't a good plan. I'd almost sooner you set a ridiculously large count...

If you're implementing a guard statement, it has the same effect as the array max size? The original while loop was just fine, it only needed a comment to say...

The code you've suggested to swap to is still several lines long and bogged down in object construction for something that already appears to be a pointer? If you truly...

> The existing machinery sometimes does not fit exactly what we want to do, which is the case here. I don't understand how it doesn't fit here. Could you clarify...

The only reason we're doing the maths to figure out the next one is because... it isn't in an array? That's precisely what arrays are there to do! If it's...

Well, ok, whatever changes you needed to achieve grater precision and coherency, go with them, but please use the array mechanism rather than manually constructing the objects? There should be...

So... I think I get what you're saying, but I fundamentally disagree that: ``` while arr[idx] and arr[idx].is_readable(): idx = idx + 1 ``` is more difficult to read than...

You're right, I hadn't seen it, but that's not using an array type at all, and it forces the target type to be a pointer but also doesn't allow the...

If we're handed the array, it should already be primed with the correct subtype so that would be all we'd need. We could also optionally accept a guard value if...