Matijn, James, Like Saikrishna, I am in favor of using the colon or pipe symbol rather than the semi-colon, for reasons that Martijn originally outlined. However, Martijn then changed his mind, which to my thinking seems a pity. Is there any chance that you could be persuaded to change it back, Martijn? A colon or a pipe symbol would make west:unsigned look like one item in a list to me, which would be handy as I'm going to return to my proposal below that on single carriageway roads we need to show the cardinal direction of OSM forward AND also OSM backward. This would create a list that could say east:unsigned;west:unsigned. Or it could say east|unsigned;west|unsigned. Personally I don't care which. But we can't use the semi-colon for both functions. I will explain why posting only OSM forward as "east" (say) can be an oversimplification further down below. But first, why the colon or pipe symbol?
My reasons are partly as you originally indicated, Martijn. Primarily, to me, a semi-colon is used in a list, such as for example the various ref tags that represent the routes that are "double banded" or "multi-banded" along a way. State DOTs often use the term "double banded" for what OSM mappers call "overlaps", in which a way carries (say) I 80;US 6 (e.g., in Iowa) or I 184;US 20;US 26 (e.g., in Boise, Idaho). Now, I think I have seen Martijn (or someone) argue that the ref tag has no purpose on the way, and should be limited to the relation, but I don't agree. In my firm's software products, made for state DOTs and public 511 traveler info systems, the first-posted route is the one the authorities (e.g., DOT; police) consider to be definitive for that way, in that it defines mile markers (*pointes kilometriques* in the metric world) and exit numbers, and is used by police and DOTs to define the route designator of the locations of crashes, etc. (Geolocation alone does not cover route designation near overpasses, or travel direction affected.) OSM mappers have suggested that these refs should be listed in highest system, lowest number order, such as I 35;I 80 or I 184;US 20;US 26, and in an ideal world DOTs would do this too; but they do not always follow this logical rule. The I 80;I 35 Des Moines northern and western bypass is posted I 80 first (I 80 determining the exit numbers and milepoints), and I changed all the ref tags accordingly from I 35;I 80 to I 80;I 35. I'm not sure how we would know which route is first-posted and therefore definitive if we did not have an ordered list in each way's ref tag, and instead used only relations. So my vote is for colons or pipe symbols to create compound directions like east:unsigned or east|unsigned. Perhaps the pipe symbol is better as it's more clearly human-read as being different from the semi-colon that I've proposed should be used to list the direction of OSM forward and then the direction of OSM backward on single carriageway roads. Note that my proposed use of the semi-colon is as part of an ordered list, in that the order of the ref tag values should match the order of the shields mounted on a vertical post, or the left-right order where shields appear on DOTs' large green signs. (Who would guess that 40 years ago I occasionally designed similar signs in the UK as a junior traffic engineer working for a county council? "Bolton A 575 Right Lane.") Back to the present day. I 80;I 35 is two data values sequentially listed, while east|unsigned is a single logical data value. Now why do I believe that we need to show the posted direction of a single carriageway road for both OSM forward and also for OSM backward? My original reason is that beltways in many states are posted east-west along their northern and southern perimeters, and north-south along their eastern and western perimeters. Ignore Washington DC's "Inner Loop", etc. I first encountered this in Minnesota where the Twin Cities beltway (I 494 and I 694) is posted east-west along the south side past the Mall of America, MSP airport and Eagan, and north-south along its eastern and western sides. I believe the north side is again posted east-west. I checked this elsewhere last week in Scottsdale, AZ, where Loop 101 is posted north-south through the Native American reservation (Pima Freeway) yet my Volvo RDS-TMC traffic info system was reporting stopped traffic eastbound on Loop 101 a few miles further north where it crosses AZ 51 and Scottsdale Road along the northern side of the loop. Now it would be nice to think that ADOT and Mn/DOT always change the posting for both directions at the same OSM node/point, but I know for a fact that this is not always so. We have found cases where a highway can be posted north in one direction and east in the other. You may well observe that most US beltways are dual carriageways, and for large cities you'd be exactly right. So if we use a different relation for each direction it's not an issue? True. But there are still some single or mixed single/dual carriageway bypasses / ring roads / beltways in the US, and even more so in other parts of the Americas (Canada; Central and South America) where I believe that AASHTO signing practices are widely emulated. I was in Argentina three years ago at the end of the Pan American Highway in Tierra del Fuego and I'm pretty sure that I saw RN 3 plated with "norte" (there is no "sur" until Buenos Aires extends RN 3 onto the Antarctic Peninsula). Its not hard to find pictures of the sign nearby that says Alaska 17,848 kilometers. When I lived in Leesburg, VA, in the late 1980s, the US 15 and VA 7 bypass was single carriageway, and (though now dual) still run in a square around 3 sides of the town. I believe that even the US must still have single carriageway bypasses and ring roads in deeply rural areas, and that somewhere on a single carraigeway, east-west must switch to north-south in a way that isn't neat and tidy. Too unusual to worry about? Perhaps. But for those of you who who care about "east|unsigned", can you imagine a state posting "east|unsigned;west" or "north|unsigned";south" on a single carriageway rural state highway? I sure can. State maintenance folks are human, and consistency is a goal never wholly achieved. I also advocate using roles of "north;south" on single carriageway roads because mixed single/dual roads are common in rural America, and they are very hard to understand (humans again) if you post the dual ways "north" and "south" and the single carriageway ways just "north" or "south" depending on the orientation of the ways' OSM forward. It is hell to check when you do this, especially in iD or Potlach. JOSM makes relations visual (or at least it will when cardinal directions are treated like forward and backward) and after a JOSM sort I can now fill in "north", "south", "north;south" and "south;north" just by staring intently at the JOSM line and arrow diagram (at least I can in the armchair sense; finding what the road crew actually signed would take a great deal longer). If you use "north;south" on single carriageways, the single carriageways leap out at you when you check the relation manually. If you use "north" and "south" to mean very different things on single carriageway ways than on dual carriageway ways, it is very hard to see whether it makes any sense or not. Not all mappers can use JOSM (I couldn't until I spent the last 3 weeks coding various mixed single/dual carriageways in ID, MN and San Diego County instead of arguing this from common sense alone) and in iD/Potlach it's a whole lot easier to see what is happening if you code single carriageways to reflect each of the two travel directions that they carry. Here I rest my case for the use of the pipe character for "east|unsigned" and the semi colon for "east;west" or "east|unsigned;west|unsigned" on single carriageways. Martijn, can I be allowed to change the wiki page once again while overeating this Christmas, should I have the time? It could be my present to myself ... Peter Davies Castle Rock Associates, Portland, OR S On Thu, Dec 19, 2013 at 10:16 PM, Saikrishna Arcot <saiarcot...@gmail.com>wrote: > I, personally, am in favor of using the colon rather than the > semicolon, for the reasons Martijn outlined a few emails back. > > Saikrishna Arcot > > On Thu 19 Dec 2013 03:23:53 AM IST, James Mast wrote: > > I have no problems going with either ":" or ";" for the separator for > > unsigned segments of highways in the role area. > > > > What does everybody else think? As this shouldn't be decided by just > > two people. We do still need the consensuses of [talk-us] before any > > mass changing of relations happen. > > > > Later tonight if I have time I'll do up an example route for this > > (US-19 Truck here in Pittsburgh) so everybody can see this in action > > at least and then we can link an example to the wiki page. The reason > > I selected the route above is because not only is it a short route, it > > does have it's middle segment hidden while on Interstates. Plus I > > have tons of experience with it having traveled it a lot in my life. > > > > -James > > > > > > _______________________________________________ > > Talk-us mailing list > > Talk-us@openstreetmap.org > > https://lists.openstreetmap.org/listinfo/talk-us > > _______________________________________________ > Talk-us mailing list > Talk-us@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-us >
_______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-us