This is reduntant. Just calculate the angle to find out where to turn to. Lulu-Ann (sorry for deleting before)
- I disagree. First of all, as described in the page, there's no way to predict what combination of left, right, and straight will be used on a highway sign based on road angles. Does I-75 go straight at I-24 because you're staying on the same highway, or does it turn because that's what the road does physically? Both are true, depending on your perspective, but only one matches the highway signs. That's not to say they couldn't have chosen the other method of signing.
- Second, I can think of places I've been where the road bends, but the most obvious instruction is "continue on [road]", or no instruction at all. Or, sometimes the road bends, and the user must be told to turn. (Contrast former and latter.) How are we supposed to deduce that just from knowing that the road curves at a certain angle? How do you distinguish between a "fork" and a "go straight/continue"? BigPeteB 05:23, 28 June 2010 (UTC)
- I agree with Lulu-Ann. The routingsoftware has to measure the lenght of following way-segments, and has to decide on this base how many segement it hast to look forward. Now it has to calculate the wohle way of the segements' angle. What's the problem. It's all to do by software. It is just a kind of way to calculate the angele of the curve oder a junction to create the right announcement. Possibly the job's also done by using the lan/longitude-Information of the nodes, for example. --Steffterra 10:43, 18 July 2010 (UTC) (so cloudmade and skobbler have to work a little bit on this, But it's redundant, like Lulu-Ann said;-)
- True, we should be able to put together a table of angles, which says "If road A and road B intersect at this angle, we should say 'Turn left', but if they intersect at that angle, we should say 'Bear left', etc." Maybe it also looks at highway=* and/or name=*. Then all the directions will be standardized.
- The problem is that as soon as you do that, I bet I can find an example on the map where the instruction we calculate from the road's angle doesn't match what people would tell you to do in real life, or what the signs on the road tell you to do. I'm all in favor of calculating it automatically when possible, but I'm certain there will be exceptions, and this proposal addresses that. --BigPeteB 15:07, 19 July 2010 (UTC)
- I strongly disagree. I could come up with plenty of examples where the geometric "road ahead" is not the logical "road ahead" and/or where it's impossible to tell by other means which way the continuation of the main road goes. It's a pity this discussion seems to have stalled as the turn=straight case is certainly one that needs to be addressed somehow. (not sure about the others) robert 00:38, 23 August 2012 (BST)