Re: [OSM-talk] Addressing Question
On Thu, 12 Nov 2009 14:29:14 -0500, Anthony o...@inbox.org wrote: On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. So your point being? These blocks can be interpolation-ways next to the way and if you like relations you can have both grouped in an associatedStreet-relation. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. First of all, how would you approximate the gap? You mean by hand? 10m along the normal of the road. Secondly, what if the houses aren't yet there? Tiger address data represents *potential* address blocks, not *actual* address blocks. There may or may not be any actual houses along those roads. Then we have to assume it's there until a mapper who can actually look for houses can correct this. That's the best we can do. Marcus ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, 2009-11-12 at 15:19 -0500, Mike N. wrote: FYI - I applied the experimental script which creates address interpolation ways at - http://svn.openstreetmap.org/applications/utils/import/tiger2osm/shape_to_osm-Tiger.py The results are at http://cid-b17e2f1a4d519b13.skydrive.live.com/self.aspx/Public/tl% 5E_2009%5E_45045%5E_addrInterpolation.zip Cool stuff! I've been looking at doing the same thing. Which osgeo python code are you using? -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
http://svn.openstreetmap.org/applications/utils/import/tiger2osm/shape_to_osm-Tiger.py Cool stuff! I've been looking at doing the same thing. Which osgeo python code are you using? I'm using the default lib for Fedora - GDAL 1.6.0; release 8.fc11 . Someone else (in Georgia?) created all the code, but I don't know if they actually uploaded the address interpolation anywhere. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [OSM-talk] Addressing Question
Ian Dees wrote: Hi everyone, I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. This has precedent already - a couple of areas in the US has Karlsruhe schema addressing converted from what is clearly centreline data to spread the addresses out on either side: http://www.openstreetmap.org/?lat=33.676517lon=-84.012017zoom=18layers=B000FTF Your point about 3 time the number of ways becomes increasing relevant with the decreasing quality of the data you are importing - 3 times the headache of fixing dreadful road geometries would be too much! On Thu, Nov 12, 2009 at 10:42 AM, Andy Robinson (blackadder-lists) ajrli...@googlemail.com wrote: I'd let someone else work out if they can transcribe the data to another format once its in OSM, should that be desirable I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! Cheers, Andy ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
Andy Allan [mailto:gravityst...@gmail.com] wrote: Sent: 12 November 2009 2:15 PM To: Andy Robinson (blackadder-lists) Cc: Ian Dees; OSM Talk; talk...@openstreetmap.org Subject: Re: [OSM-talk] Addressing Question Ian Dees wrote: Hi everyone, I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. This has precedent already - a couple of areas in the US has Karlsruhe schema addressing converted from what is clearly centreline data to spread the addresses out on either side: http://www.openstreetmap.org/?lat=33.676517lon=- 84.012017zoom=18layers=B000FTF Your point about 3 time the number of ways becomes increasing relevant with the decreasing quality of the data you are importing - 3 times the headache of fixing dreadful road geometries would be too much! On Thu, Nov 12, 2009 at 10:42 AM, Andy Robinson (blackadder-lists) ajrli...@googlemail.com wrote: I'd let someone else work out if they can transcribe the data to another format once its in OSM, should that be desirable I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! You make a good point and I certainly wouldn't want to see data imported that was either difficult to rework or didn't make logical sense. If its been done before to offset left/right data automagically I too would vote for that as a preferred import method. Cheers Andy R ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
Which somehow reminds of the AND data imported to OSM. I am not sure whether AND and TIGER had anything to do with each other but most of the highways from the AND data in India are straight lines, often a couple of hundred metres off the actual road. I have been deleting old tracks and adding new tracks. Just to give you an example, refer the below frame. http://www.openstreetmap.org/?lat=32.1488lon=76.4172zoom=13layers=B000FTF This road, a curvy mountain road was a straight line even at the highest zoom level with no waypoints whatsoever till 2-3 days ago. I have deleted 3 such roads in the last week and replaced with the new ones. And I know there are thousands more to go. I doubt if roads like these actually add any value, either from a mapping or routing point of view to OSM. Regards, Shalabh On Thu, Nov 12, 2009 at 10:47 PM, Andy Robinson (blackadder-lists) ajrli...@googlemail.com wrote: Andy Allan [mailto:gravityst...@gmail.com] wrote: Sent: 12 November 2009 2:15 PM To: Andy Robinson (blackadder-lists) Cc: Ian Dees; OSM Talk; talk...@openstreetmap.org Subject: Re: [OSM-talk] Addressing Question Ian Dees wrote: Hi everyone, I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. This has precedent already - a couple of areas in the US has Karlsruhe schema addressing converted from what is clearly centreline data to spread the addresses out on either side: http://www.openstreetmap.org/?lat=33.676517lon=- 84.012017zoom=18layers=B000FTF Your point about 3 time the number of ways becomes increasing relevant with the decreasing quality of the data you are importing - 3 times the headache of fixing dreadful road geometries would be too much! On Thu, Nov 12, 2009 at 10:42 AM, Andy Robinson (blackadder-lists) ajrli...@googlemail.com wrote: I'd let someone else work out if they can transcribe the data to another format once its in OSM, should that be desirable I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! You make a good point and I certainly wouldn't want to see data imported that was either difficult to rework or didn't make logical sense. If its been done before to offset left/right data automagically I too would vote for that as a preferred import method. Cheers Andy R ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. First of all, how would you approximate the gap? You mean by hand? Secondly, what if the houses aren't yet there? Tiger address data represents *potential* address blocks, not *actual* address blocks. There may or may not be any actual houses along those roads. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
2009/11/12 Anthony o...@inbox.org: On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. If we're going to go into detail, no type of interpolation reflects reality, it's just interpolation. It's there to help the mappers add proper address / other information and in the meantime give users approximate geocoding functionality. If you look at TIGER a lot of it doesn't reflect reality, it just one step. Cheers ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
FYI - I applied the experimental script which creates address interpolation ways at - http://svn.openstreetmap.org/applications/utils/import/tiger2osm/shape_to_osm-Tiger.py The results are at http://cid-b17e2f1a4d519b13.skydrive.live.com/self.aspx/Public/tl%5E_2009%5E_45045%5E_addrInterpolation.zip (7 MB zipped, 88 MB unzipped). Runtime for my county was about 45 seconds.___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [Talk-us] [OSM-talk] Addressing Question
On 12 Nov 2009, at 8:28 , Ian Dees wrote: On Thu, Nov 12, 2009 at 10:20 AM, Apollinaris Schoell ascho...@gmail.com wrote: On 12 Nov 2009, at 6:14 , Andy Allan wrote: I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! +10 Give everyone a chance to work in a constructive way and don't expect others to clean the mess bad import left behind. No wonder there are only few motivated mappers in US. In Canada they do a much better job in integrating the community and don't import every shape file blindly just because it's available. Not to start a holy war over what community is better, but there weren't a whole lot of mappers in the US (at least on talk or talk- us) when we started doing the TIGER import… It's not about better community. The point is an empty map doesn't attract average mappers. someone ambitious needs to start the whole thing. A nearly finished map doesn't attract the ambitious mappers. A broken map doesn't attract either one. Imports help with the first challenge. Bad imports not at all. Tiger import was very useful and done in the best way known at that time. but has many problems. technical and community wise. We shouldn't make the same mistakes again just to fill the database. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, 2009-11-12 at 10:28 -0600, Ian Dees wrote: Give everyone a chance to work in a constructive way and don't expect others to clean the mess bad import left behind. No wonder there are only few motivated mappers in US. In Canada they do a much better job in integrating the community and don't import every shape file blindly just because it's available. Not to start a holy war over what community is better, but there weren't a whole lot of mappers in the US (at least on talk or talk-us) when we started doing the TIGER import... People were being actively told not to map in the US because we had TIGER coming and it would replace any work you ended up doing. The standard OSM user tries to find their street first. The typical US OSM experience has gone from, My street isn't there to My street is crooked. -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 10:20 AM, Apollinaris Schoell ascho...@gmail.comwrote: On 12 Nov 2009, at 6:14 , Andy Allan wrote: I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! +10 Give everyone a chance to work in a constructive way and don't expect others to clean the mess bad import left behind. No wonder there are only few motivated mappers in US. In Canada they do a much better job in integrating the community and don't import every shape file blindly just because it's available. Not to start a holy war over what community is better, but there weren't a whole lot of mappers in the US (at least on talk or talk-us) when we started doing the TIGER import... ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On 12 Nov 2009, at 6:14 , Andy Allan wrote: I disagree there. It's much better to put the effort in during the initial import, than to import things badly and try to fix it up later. We've been working on lots of post-import fixups in the last 6 months and it's much harder than everyone assumes. The 4 months to remove TIGER node tags is a case in point - it took less time than that to import them! +10 Give everyone a chance to work in a constructive way and don't expect others to clean the mess bad import left behind. No wonder there are only few motivated mappers in US. In Canada they do a much better job in integrating the community and don't import every shape file blindly just because it's available. Cheers, Andy ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 11:17 PM, Apollinaris Schoell ascho...@gmail.com wrote: On Wed, Nov 11, 2009 at 7:19 PM, Anthony o...@inbox.org wrote: It probably has to be a relation. Include a start node, an end node, and a list of one or more ways (which are connected to form one logical way). the ways have to be split at the start/end node. Not if you use a relation. the relation members have to be ordered. No they don't. how is that easier than the Karlsruhe scheme? It's not really easier so much as more correct. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 11:26 PM, Ian Dees ian.d...@gmail.com wrote: On Wed, Nov 11, 2009 at 10:17 PM, Apollinaris Schoell ascho...@gmail.com wrote: On Wed, Nov 11, 2009 at 7:19 PM, Anthony o...@inbox.org wrote: On the other hand, putting the information directly on the way would be problematic for many reasons. Ranges might span multiple ways, and right/left has to be reversed whenever the way is reversed being the most troublesome. this is enough reason to stay away from such a scheme. if it's too difficult no one will use it or they will break the data. This scheme works for all of the places that I'm sourcing data from... they have line segments that are tagged with the left/right-begin/end addresses. Each road is broken up into line segments that have different address values. I'm not sure what your data is like, but the Tiger data inaccurately splits the address ranges when it needs to split a segment. In other words, if a road goes from 2 to 100, and it needs to be split in half, Tiger blindly splits the segments up as 2 - 48 and 50-100 *without even checking if this is correct*. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. I'd say it's better to approximate the gap between the road and the houses (10m?) than to just put it on the centreline due to that being easier. First of all, how would you approximate the gap? You mean by hand? Secondly, what if the houses aren't yet there? Tiger address data represents *potential* address blocks, not *actual* address blocks. There may or may not be any actual houses along those roads. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On 12 Nov 2009, at 11:29 , Anthony wrote: On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. Don't know any place except in US where this has been done. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, 2009-11-12 at 11:40 -0800, Apollinaris Schoell wrote: On 12 Nov 2009, at 11:29 , Anthony wrote: On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. Don't know any place except in US where this has been done. So, should we ignore the US for addressing entirely since it is different? Or, should US addressing use a different scheme than the rest of the world? We like being different. -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 2:40 PM, Apollinaris Schoell ascho...@gmail.com wrote: Don't know any place except in US where this has been done. Even if it weren't done anywhere else (which it is, see below), there are a lot of houses in the US. how is that easier than the Karlsruhe scheme? It's not really easier so much as more correct. why more correct? Because arbitrarily locating a way 10 meters (or whatever) away from the road centerline adds artificial precision. the address is an attribute of the house not an attribute of the street. If you want to get technical an actual address is usually an attribute of a mail delivery point, which may or may not correspond to a house. But potential addresses (which are likely the data which we are contemplating importing) are an attribute of the street. If you want to just leave the data out of OSM altogether, fine. That might be a good idea. Geocoders can use the data within the system it was designed for. But if you're going to import it into OSM, you shouldn't add artificial precision to it. On Thu, Nov 12, 2009 at 2:41 PM, andrzej zaborowski balr...@gmail.com wrote: If we're going to go into detail, no type of interpolation reflects reality, it's just interpolation. I disagree. An approximation of reality reflects reality. On Thu, Nov 12, 2009 at 2:46 PM, Liz ed...@billiau.net wrote: On Fri, 13 Nov 2009, Apollinaris Schoell wrote: Don't know any place except in US where this has been done. The rural numbering scheme used in Australia is an excellent example, as the potential number of addresses is the length of the road in km * 100, and there is no expectation that any number of the addresses will be used. I'm curious: Does Australia use left/right numbering? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
follow the OSM principle. map what's on the ground no matter where you are On 12 Nov 2009, at 11:56 , Dave Hansen wrote: On Thu, 2009-11-12 at 11:40 -0800, Apollinaris Schoell wrote: On 12 Nov 2009, at 11:29 , Anthony wrote: On Thu, Nov 12, 2009 at 9:14 AM, Andy Allan gravityst...@gmail.com wrote: It's a fairly well established convention that in OSM it's the houses/plots, not the road centrelines, that are addressed. But that doesn't always reflect reality. The reality, at least in many parts of the world, is that the streets are given blocks of potential addresses, and the houses/plots/whatever are given actual addresses from those potential address blocks. Don't know any place except in US where this has been done. So, should we ignore the US for addressing entirely since it is different? Or, should US addressing use a different scheme than the rest of the world? We like being different. -- Dave ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
If we're going to go into detail, no type of interpolation reflects reality, it's just interpolation. I disagree. An approximation of reality reflects reality. Physical street surveys will almost never get 100% reality due to missing house numbers, etc. Are you proposing to discourage physical street surveys that are not 100% complete just because the data is all not there? No, just the opposite. On Thu, Nov 12, 2009 at 3:06 PM, Apollinaris Schoell ascho...@gmail.com wrote: follow the OSM principle. map what's on the ground no matter where you are What's on the ground changes from place to place: http://en.wikipedia.org/wiki/House_numbering In countries like Brazil and Argentina, but also in some villages in France, this scheme is used also for streets in cities, where the house number is the distance, measured in meters, from the house to the start of the street. For people living near highways or roads [in Latin America] the usual address is the kilometer of the road in which the house is established [...] In semi-rural and rural areas [of Australia], where houses and farms are widely spaced, a numbering system based on tens of metres or (less commonly) metres has been devised. Thus a farm 2300m from the start of the road, on the right-hand side would be numbered 230. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
FYI - I applied the experimental script which creates address interpolation ways at - http://svn.openstreetmap.org/applications/utils/import/tiger2osm/shape_to_osm-Tiger.py The results are at http://cid-b17e2f1a4d519b13.skydrive.live.com/self.aspx/Public/tl%5E_2009%5E_45045%5E_addrInterpolation.zip (7 MB zipped, 88 MB unzipped). Runtime for my county was about 45 seconds.___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 4:43 PM, andrzej zaborowski balr...@gmail.com wrote: Every single country has different addressing rules, it's not like this particular scheme is special. That's why someone came up with a tagging scheme that can express all or most of these rules They did? What scheme is that? My understanding of the history of the Karlsruhe Schema is that a bunch of people got together in Karlsruhe and came up with something that worked for them, explicitly stating that other people should develop something that works for them. If we're going to have special tagging for US then let's have it for every country and let's split into many projects and just give up on this whole silly idea of having a single map. Is there a reason you're ignoring the fact that potential addressing is used in many places outside the US? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 5:14 PM, andrzej zaborowski balr...@gmail.com wrote: Ian hasn't (yet) mentioned whether this data he deals with contains potential address ranges or actual ranges, so I assumed actual. The fact that it's tagged on the line segments representing the road centerline pretty much guarantees that it's potential. I highly doubt they're splitting the line segment every single time a number gets skipped. I agree it may be useful to have the potential assigned range in the db, too, using whatever tagging (or in a separate db, since this is not stuff on the ground). I can see keeping it in a separate db, and really I'm leaning toward that as being the best option. What are the advantages of having this in the OSM db? When the roads change, you're going to have to either re-survey the data or throw out the address ranges anyway. The address ranges are pretty much only useful within the context of the original road centerlines. Geocoding or reverse-geocoding software can connect between the two databases using latitude/longitude pairs. I can't really see any point in integrating it. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 5:32 PM, Mike N. nice...@att.net wrote: There are cases with Karlruhe Scheme that need addditional tags like Czech addresses but I haven't heard of such cases from US or other mappers. I recently started using a new modifier tag addr:inclusion to help in accurately tagging my survey data (and added it to the wiki http://wiki.openstreetmap.org/wiki/Key:addr ) Cool. I invented that tag, and that's pretty close to what I meant by it. :) ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
Hi, Anthony wrote: On Thu, Nov 12, 2009 at 4:43 PM, andrzej zaborowski balr...@gmail.com wrote: Every single country has different addressing rules, it's not like this particular scheme is special. That's why someone came up with a tagging scheme that can express all or most of these rules They did? What scheme is that? My understanding of the history of the Karlsruhe Schema is that a bunch of people got together in Karlsruhe and came up with something that worked for them, explicitly stating that other people should develop something that works for them. Speaking as one of that bunch of people; that ist correct. Of course when developing the Karlsruhe Schema we hoped that our work would be usable for others as well, but we were under no illusion that there are places where it doesn't. (Not sure if I got the double negative right in that sentence but you'll know what I mean.) To be honest, at the time we thought that those places would be in the less developed world, where sometimes houses are reported to have addresses like 3rd house on left hand side after the tree that looks like an elephant. We didn't think that the US would be so different as to require their own schema. If you do develop your own schema then I would ask you to consider to at least adhere to the following basic idea that we used: We said that in the long run, we expect every single house to be on the map - either as a node or, more likely, as a building outline - and carry its own number. Interpolation ranges, therefore, were meant to be something easy for situations where you cannot be bothered to do it right. Our expectation is that in the long run, interpolation lines will be obsolete. An interpolation line still maps more or less what's on the ground - at least the house numbers at both ends of the line will have been surveyed. Many people in Germany even break their interpolation lines if houses are missing in between, i.e. if a road has the house numbers 100, 102, 104, 108, 110, 112 they will create one interpolation line for 100-104 and one for 108-112. Now if I understand your situation correctly, the only difference you have is that your interpolation lines are one step more abstract; they don't give a range of numbers of definitely existing addresses, but instead give the range of valid numbers in the block concerned. So you know that *if* there is a house #1300 it will be there (but it might not exist at all). My suggestion to this would be to use something like our interpolation lines but give them a different name (e.g. instead of addr:interpolation call it addr:range or something). I would also strongly encourage you to use one such line on each side of the road, instead of putting tags on the road itself. This makes it very clear which side an address is on, better than any tags you can put on the way, no matter how many left/right prefixes or suffixes you add to those tags. This is one thing we discussed at length when setting up the Karlsruhe schema; even here, many people advocated putting something like left:from=15, left:to=25, right:from=12, right:to=24 on the ways but we'd have none of that. One of many reasons for that being that this would interfere with ways being split or combined - this must be doubly true for your schema: If you have to split a road in the middle of a block because a speed limit starts there, how will you know which theoretical house number would be at the split if the house hasn't even been built yet? Of course, and that's the but, if such a schema leads to millions of address range ways in places where no houses have been built, then that's perhaps a bit confusing... Thanks for listening. Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 5:41 PM, Frederik Ramm frede...@remote.org wrote: I would also strongly encourage you to use one such line on each side of the road, instead of putting tags on the road itself. This makes it very clear which side an address is on, better than any tags you can put on the way, no matter how many left/right prefixes or suffixes you add to those tags. Would you map a no right turn as a node 7 meters behind and to the right of an intersection? After all, that's more or less what's on the ground. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
Hi, Anthony wrote: On Thu, Nov 12, 2009 at 5:41 PM, Frederik Ramm frede...@remote.org wrote: I would also strongly encourage you to use one such line on each side of the road, instead of putting tags on the road itself. This makes it very clear which side an address is on, better than any tags you can put on the way, no matter how many left/right prefixes or suffixes you add to those tags. Would you map a no right turn as a node 7 meters behind and to the right of an intersection? After all, that's more or less what's on the ground. I might be missing some irony here. I don't know the significance of no right turn for you, and I don't know what it has to do with addressing. Traffic signs, at least where I live, usually are there as a physical reminder (or notification) of an abstract concept. The administration makes a certain decision - for example, that parking should not be allowed in a certain location, or a speed limit should be put in place, or whatever. Then signs are put up to inform people of this decision. The exact location of the signs is often an implementation detail. The sign itself is irrelevant; the abstract concept is what matters. I try to map the abstract concept wherever possible. Consequently, I'd map a no right turn as a relation involving two ways, and not in the form of a traffic sign. I consider interpolation ways to be an abstract thing also. To convey the information, they need to be on each side of the road, but, if that was your question, to me it doesn't matter whether they are a few centimetres away from the road or 10 metres away. As long as there is no doubt (for the person viewing the situation in an editor) which road they belong to, it's fine. In practice it turns out that you often draw the lines approximately where the houses would be on the ground, but to me that is not relevant. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 6:43 PM, Frederik Ramm frede...@remote.org wrote: I consider interpolation ways to be an abstract thing also. To convey the information, they need to be on each side of the road The thing is, they don't. As long as there is no doubt (for the person viewing the situation in an editor) which road they belong to, it's fine. You mean with the addr:street tag? Other than that tag, it's quite common (at least here in Florida) for it to be non-obvious which road an address belongs to, if you want to put the interpolation way over top of the houses. The building might very well be far away from the road it belongs to, and closer to another road that it doesn't belong to. For geolocation using actual addresses, that's a feature, not a bug - but it doesn't work for potential address ranges at all. On Thu, Nov 12, 2009 at 6:56 PM, andrzej zaborowski balr...@gmail.com wrote: 2009/11/12 Anthony o...@inbox.org: I can see keeping it in a separate db, and really I'm leaning toward that as being the best option. What are the advantages of having this in the OSM db? When the roads change, you're going to have to either re-survey the data or throw out the address ranges anyway. The address ranges are pretty much only useful within the context of the original road centerlines. Geocoding or reverse-geocoding software can connect between the two databases using latitude/longitude pairs. I can't really see any point in integrating it. On the other hand if you receive a data donation from a densely built-up city's council then there will be more existing addresses than non-existing ones in the area and it will be easier for local mappers to start with all the possible interpolation ways and slowly remove fragments than to survey with an empty map. And will better match reality. *Nod*. If you can at least semi-manually integrate the data so you're pretty sure maybe 95% of it is in the general vicinity of the buildings, that's cool. But that's only going to be feasible in certain locations. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [OSM-talk] Addressing Question
Hiya, 2009/11/12 Ian Dees ian.d...@gmail.com: I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. The hope is that local mappers there will be slowly improving imported data until there are separate points for every address I think? Then I'd recommend just adding those separate ways and making it easier for the mappers to build on this data, instead of making it harder for nearly everyone dealing with OSM data by adding a whole different addressing scheme. Cheers ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 7:55 PM, andrzej zaborowski balr...@gmail.comwrote: Hiya, 2009/11/12 Ian Dees ian.d...@gmail.com: I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. The hope is that local mappers there will be slowly improving imported data until there are separate points for every address I think? Then I'd recommend just adding those separate ways and making it easier for the mappers to build on this data, instead of making it harder for nearly everyone dealing with OSM data by adding a whole different addressing scheme. No, I doubt local mappers will improve the data. I sent this mail because almost all of the data I've seen available for import in the US (all the way from individual municipalities to TIGER's shapefiles) has this left/right-from/to scheme for addressing information. If the expectation is that we will always be following the Karlsruhe schema (with separate ways on each side of the road centerline), then importing this addressing data will be next to impossible*. -Ian * Ok, not impossible, but the import size would triple and the CPU time to compute the new addressing-only ways might make it hard for the regular mapper to do. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 9:47 PM, andrzej zaborowski balr...@gmail.comwrote: That's a pretty pessimistic view. Sorry, I am pretty grumpy today. The area I'm looking at actually has quite a few mappers already, so I imagine this data would probably get updated quickly. For the record an import I've done had only this left/right - to/from housenumber information, too, so I have an ugly python script here ready to throw at this kind of data (after adapting to whatever the input format is) and I would be happy to do the processing on my PC if you decide to go this way. The whole toolchain should still behave reasonably for data size of TIGER (though obviously I didn't have that much data) What does the math look like to handle intersections? Curvy roads? ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
2009/11/12 Ian Dees ian.d...@gmail.com: What does the math look like to handle intersections? Curvy roads? I admit that data wasn't complex, all segments were straight and all nodes were treated as intersections. Do you have parametric (e.g. bezier) curves or just lots of short segments making soft turns? (I'm optimistic we can figure something out in both cases) Regards ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
2009/11/12 Ian Dees ian.d...@gmail.com: On Wed, Nov 11, 2009 at 7:55 PM, andrzej zaborowski balr...@gmail.com wrote: The hope is that local mappers there will be slowly improving imported data until there are separate points for every address I think? Then I'd recommend just adding those separate ways and making it easier for the mappers to build on this data, instead of making it harder for nearly everyone dealing with OSM data by adding a whole different addressing scheme. No, I doubt local mappers will improve the data. I sent this mail because That's a pretty pessimistic view. almost all of the data I've seen available for import in the US (all the way from individual municipalities to TIGER's shapefiles) has this left/right-from/to scheme for addressing information. If the expectation is that we will always be following the Karlsruhe schema (with separate ways on each side of the road centerline), then importing this addressing data will be next to impossible*. This would mean we are excluding some features, or some tagging schemes, in some parts of the world due to disk space or processing time but not in other parts of the world. I'm sure we will never have uniform tagging and uniform data quality everywhere but I still want to aim at it. For the record an import I've done had only this left/right - to/from housenumber information, too, so I have an ugly python script here ready to throw at this kind of data (after adapting to whatever the input format is) and I would be happy to do the processing on my PC if you decide to go this way. The whole toolchain should still behave reasonably for data size of TIGER (though obviously I didn't have that much data) Cheers ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Addressing Question
On Thu, Nov 12, 2009 at 2:13 AM, Ian Dees ian.d...@gmail.com wrote: Hi everyone, I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Can you identify the location of the Addr for the From/To? If so it would be easy to calculate interpolation-ways right and left of the streets using the Karlsruhe Schema. Is there an accepted way of applying these tags to the road ways? No, only in applying it to houses next to the street and special ways connecting next to the street for interpolating whole streets of house-numbers at once. It doesn't really make very much sense to create and store a separate way just for the addressing information. I disagree here because of the hundrets of special cases that absolutely must be handled to be correct that come from the fact that houses are not usually build in the middle of a road. ...but I might have arrived too late in the argument to say that :-)... Yes, you are. ;) PS: I`m not reading the talk-US but as you crossposted there, so do I. Please let me know of yet another addressing-schema comes that is in actual usage to make sure my address-search does work on all the planet. Marcus ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 7:55 PM, andrzej zaborowski balr...@gmail.comwrote: Hiya, 2009/11/12 Ian Dees ian.d...@gmail.com: I'm looking at some donated street centerline data that has addressing data in the form of Right/Left From Addr and Right/Left To Addr on each street centerline. Is there an accepted way of applying these tags to the road ways? It doesn't really make very much sense to create and store a separate way just for the addressing information. The hope is that local mappers there will be slowly improving imported data until there are separate points for every address I think? Then I'd recommend just adding those separate ways and making it easier for the mappers to build on this data, instead of making it harder for nearly everyone dealing with OSM data by adding a whole different addressing scheme. No, I doubt local mappers will improve the data. I sent this mail because almost all of the data I've seen available for import in the US (all the way from individual municipalities to TIGER's shapefiles) has this left/right-from/to scheme for addressing information. If the expectation is that we will always be following the Karlsruhe schema (with separate ways on each side of the road centerline), then importing this addressing data will be next to impossible*. -Ian * Ok, not impossible, but the import size would triple and the CPU time to compute the new addressing-only ways might make it hard for the regular mapper to do. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 9:21 PM, Ian Dees ian.d...@gmail.com wrote: No, I doubt local mappers will improve the data. If that's true (and I'm really not sure if it is), then it really shouldn't be in OSM in the first place. I sent this mail because almost all of the data I've seen available for import in the US (all the way from individual municipalities to TIGER's shapefiles) has this left/right-from/to scheme for addressing information. If the expectation is that we will always be following the Karlsruhe schema (with separate ways on each side of the road centerline), then importing this addressing data will be next to impossible*. On the other hand, putting the information directly on the way would be problematic for many reasons. Ranges might span multiple ways, and right/left has to be reversed whenever the way is reversed being the most troublesome. It probably has to be a relation. Include a start node, an end node, and a list of one or more ways (which are connected to form one logical way). ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
Ian Dees wrote: * Ok, not impossible, but the import size would triple and the CPU time to compute the new addressing-only ways might make it hard for the regular mapper to do. But for no added code and editor complexity. IMHO the only decent alternative is using a relation for each address interpolation-range, with the nodes at the ends of the range and the way itself as members. This is roughly half as expensive as the Karlsruhe interpolation ways, and reduces the visual clutter in the editors. It is also insensitive to way direction reversal. One of the worst problems is when you split a way which has more than one such relation (or even one, if the split is NOT between the two nodes). Consider a way A from node 1 to 2 to 3; interpolation relation R has members 1[from], 2[to], A[via] and the range info. Relation S has members 2[from], 3[to], A[via]. A is split into A and B at node 2. If B gets all tags and relations from A, things get ugly - R and S now have 2 via members. So the editor needs to know how to remove the extra relations so that only ways that contain both the nodes stay members. Note: This problem already exists in reverse for turn restrictions. Both the from and to ways are required to end at the via node. So, say, Market St crosses 1st St. Market and 1st are both split at the intersection so a no_left_turn can be inserted. If some well-meaning editor goes and rejoins the two halves of Market, the relation is at least formally wrong. /Stellan ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 7:19 PM, Anthony o...@inbox.org wrote: On the other hand, putting the information directly on the way would be problematic for many reasons. Ranges might span multiple ways, and right/left has to be reversed whenever the way is reversed being the most troublesome. this is enough reason to stay away from such a scheme. if it's too difficult no one will use it or they will break the data. It probably has to be a relation. Include a start node, an end node, and a list of one or more ways (which are connected to form one logical way). the ways have to be split at the start/end node. the relation members have to be ordered. too many beginners and medium experienced mappers have problems to understand such a scheme. how is that easier than the Karlsruhe scheme? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
Ian Dees wrote: * Ok, not impossible, but the import size would triple and the CPU time to compute the new addressing-only ways might make it hard for the regular mapper to do. But for no added code and editor complexity. IMHO the only decent alternative is using a relation for each address interpolation-range, with the nodes at the ends of the range and the way itself as members. This is roughly half as expensive as the Karlsruhe interpolation ways, and reduces the visual clutter in the editors. It is also insensitive to way direction reversal. One of the worst problems is when you split a way which has more than one such relation (or even one, if the split is NOT between the two nodes). Consider a way A from node 1 to 2 to 3; interpolation relation R has members 1[from], 2[to], A[via] and the range info. Relation S has members 2[from], 3[to], A[via]. A is split into A and B at node 2. If B gets all tags and relations from A, things get ugly - R and S now have 2 via members. So the editor needs to know how to remove the extra relations so that only ways that contain both the nodes stay members. Note: This problem already exists in reverse for turn restrictions. Both the from and to ways are required to end at the via node. So, say, Market St crosses 1st St. Market and 1st are both split at the intersection so a no_left_turn can be inserted. If some well-meaning editor goes and rejoins the two halves of Market, the relation is at least formally wrong. /Stellan ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] [OSM-talk] Addressing Question
On Wed, Nov 11, 2009 at 9:47 PM, andrzej zaborowski balr...@gmail.comwrote: That's a pretty pessimistic view. Sorry, I am pretty grumpy today. The area I'm looking at actually has quite a few mappers already, so I imagine this data would probably get updated quickly. For the record an import I've done had only this left/right - to/from housenumber information, too, so I have an ugly python script here ready to throw at this kind of data (after adapting to whatever the input format is) and I would be happy to do the processing on my PC if you decide to go this way. The whole toolchain should still behave reasonably for data size of TIGER (though obviously I didn't have that much data) What does the math look like to handle intersections? Curvy roads? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
