sbelekevich wrote:Are there that many of these intersections out there? I don't think I've never been through one.

Lots, mostly in cities, and they're easy to draw badly.

WeeeZer14 wrote:Would it be a good idea to include links to wikipedia in the wiki? There are some reasonable descriptions and real-world pictures of many of the interchange types that could be helpful.

Absolutely. They appear elsewhere, so why not.

WeeeZer14 wrote:I added Folded Diamonds and Cloverleafs. I listed two ways to handle part of the Cloverleaf. Does that make sense to give options or should we just stick with one way?

You explain why the one-node situation is helpful and show how to avoid potential problems with drawing it that way, so I don't see any reason to change.

Re the angle in the entrance ramps: yes, not required, but not implied in the text as necessary, either (right?).

When the angle on the exit ramps is done right, it will be invisible anyway. The illustrations make it clear that it needs to be there, but only at high zoom levels. Would screenshots of the client help?

I can't think of anything else absolutely necessary. Collector-distributor lanes seem to be losing popularity, and most of the innovative interchanges like double-crossover diamond haven't caught on widely yet, and probably should be avoided in order to reduce potential for confusion.

CBenson wrote:...some of the color schemes really make the ramps look like a gap in the road.

One way to deal with this is to make the ramp segments a lot shorter. The ones in the Milwaukee example are 100m long...they could be 10m long and still work.

Very nice, but I do have one comment: You don't need to add a geometry node to a ramp exit in order to make the turn permissions visible. Simply hit the 's' key in the WME and hidden permission indicators slide out from underneath the ones hiding them.

A DDI section would definitely be helpful as the first one in Georgia opened recently and more than a few are scheduled to be built soon

I-285 at Ashford Dunwoody Rd (does there need to be a junction at the crossover point?):
https://www.waze.com/editor/?zoom=5&lat ... FTTFTTTTFT

As a noob I just want to say how lovely this page is. Clear text, lots of images, great organization. More!

sbelekevich wrote:For way finder segments, an editor in milwaukee did this: https://www.waze.com/editor/?zoom=6&lat ... FFTFTTTTFT. It looks a little goofy in the client, but routing instructions work really well, and you get all the info from the pathfinder signs without having the text show up on the client map. You are also guaranteed an instruction. Any thoughts on this practice? (it is also used in chicago)

Wouldn't a topology kink in I-94 E at the junction do the same?

CBenson wrote:

jwkilgore wrote:Also a few people (whose opinions I value) state that it doesn't need to be done, but no actual reasons as to why not.

Well I'll restate my reason why not. I don't pan around the map when I'm driving. If I'm about to enter a miles long segment that has a slow down at the other end, the only way I'm going to know about it is if the entire segment is marked as slow. If you split the segment up so that the slow down is marked only on segment that is not yet on my display, it's of no help to me.

This is an important point. Here's one for the other side of the argument: suppose I see a section of road marked orange. I start in and there are no delays so I figure Waze is showing me old information ... until suddenly I have to slam on the brakes for stopped traffic. It would be safer for me if I had fine-grained information. Here's another special case: a two-way street with my destination in the middle. If I can see that it's red on one end but not on the other, I can go around (most of) the red. But if I had to choose right now I would delete all extraneous junctions and tell people to drive safely.

WeeeZer14 wrote:My point is that the granularity may give a false sense of accuracy that doesn't really exist. It is accuracy vs precision. We don't need to be more accurate than we are precise.

Persuasive.