I see now, that's good. The way refs should still probably only be CA 99 though, especially in this case where the possible Interstate designations are not really very far along.

Hi,

This probably shouldn't go in ref=*. fut_ref=* on the ways should be fine, but the ideal way to do this would be a route relation with network=US:I:Future

Also see discussion here: https://community.openstreetmap.org/t/planned-edit-to-re-tag-descriptive-names-in-road-route-relations-in-the-united-states/108454

Hello,

I see that you moved "New York State Route 9A" from the description=* tag to the name=* tag. For road routes, there is some amount of agreement that name=* shouldn't be used like this, since it is just a description of what is in the ref=* tag. See more here (the "Interstate 5" example): osm.wiki/Names#Name_is_the_name_only

Hello,

Please be sure to only map turn restrictions when they are actually signed on the ground. Some of the restrictions you've created look like you just figured that cars shouldn't do that, but this is usually not enough to add a turn restriction.

Generally, in cases like the I-91/CT-159 interchange, I feel that it is better to not change the classification of the connecting road (Meadow Road in this case). The ground truth is that CT-159 is the primary road, and you have to get off onto a side street to access it. If we tag the side street as primary, this information is lost, or at least obscured. I would be more okay with giving highway=tertiary to this particular segment, but if it was really a residential street with houses on either side, I think it should stay as highway=residential.

In contrast, I am okay with tagging the side street higher if it is the beginning/end of a route of that classification. For example, this segment connecting I-84 to the primary route on CT-34 should stay as-is: way/411203807 . This is especially the case if a numbered route of that classification is signed along the ramp and side street.

I don't know if I agree with highway implying line geometry; I think it's just usually true. In this case, highway=rest_area is only supposed to be used on nodes and closed ways, so it seems kind of redundant to include area=yes. Data consumers should just refrain from using objects with tags that they don't understand, even if they have a general rule for the key (i.e. don't treat a rest area as a routable highway)