Re: [OSM-talk] address interpolation
On Wed, Sep 30, 2009 at 10:24:06AM +0100, David Earl wrote: And I think the previous point about other alphabets is a red herring too: basically no letter can precede all other letters in whatever alphabet. If there's ambiguity in the alphabet due to the glyph The problem is as follows: You see an interpolation 25a to 25c. How do you know that this means 25a, 25b, 25c? You know by removing the number and then starting with the a go through code points adding one until you reach c. Easy. This will work for all alphabets where that are layed out in alphabetical order in Unicode, and they probably all are. (but thats an assumption on my part :-) But, when you see an interpolation 25 to 25c. How do you know that this means 25, 25a, 25b, 25c? You again remove the number. That gives you the first entry. But what happens then? You have to look at the c, decide that thats a latin alphabet, the first letter of that is a and you go on with that. So you have to know all letters in all alphabets in the world to make this algorithm work. It might not be a big deal in practice, just another case where our assumptions might not hold everywhere. Jochen -- Jochen Topf joc...@remote.org http://www.remote.org/jochen/ +49-721-388298 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
The problem is as follows: You see an interpolation 25a to 25c. How do you know that this means 25a, 25b, 25c? You know by removing the number and then starting with the a go through code points adding one until you reach c. Easy. This will work for all alphabets where that are layed out in alphabetical order in Unicode, and they probably all are. (but thats an assumption on my part :-) But, when you see an interpolation 25 to 25c. How do you know that this means 25, 25a, 25b, 25c? You again remove the number. That gives you the first entry. But what happens then? You count to the end of this interpolation-way (there are 4 nodes in, 1 is the start and 1 is the end, so 2 in between) and substract this number (2) from the last char (c). Following your assumptions from above this should work (when you can add 1 to go to the next char, you can also subtract 1 to go to the previous). This way you'll first get b and then a - those are the chars for your houses. Peter ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
The problem is as follows: You see an interpolation 25a to 25c. How do you know that this means 25a, 25b, 25c? You know by removing the number and then starting with the a go through code points adding one until you reach c. Easy. This will work for all alphabets where that are layed out in alphabetical order in Unicode, and they probably all are. (but thats an assumption on my part :-) Ouch. Unicode order has no meaning in the real world, and only really works for English (and not even then properly for subtle cases, like ligatures, not that these would ever be used in these kind of addresses). You need to know the lexical ordering, which means you need to know the language. Sometimes you can guess from the character, and two characters make it easier than one, but the problem doesn't go away with two - the null variant isn't central to this problem. There's also a cultural assumption about how you might do this in other countries. I've no idea how Chinese addresses are formulated normally - whether they even use digits, and if those digits are the arabic numerals - let alone what these exceptional cases might be. But IF you know it is Chinese and IF the scheme fits, with digits + Chinese Character, then the null case still works (Chinese characters still have a lexical ordering, I believe it has to do with the number of strokes, but any relationship to Unicode order is purely coincidental) So I'm coming round to the view that alphabetic should explicitly only mean only n nA nB ... nZ where you can start and end at any point in the sequence, and not even try to deal with other characters from other alphabets (not even other latin ones). Any other sequence from other cultures needs its own interpolation style or additional qualifying tag to identify it, just as we'd tag an email with the encoding. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On Fri, Oct 02, 2009 at 10:26:11AM +0200, Peter Körner wrote: The problem is as follows: You see an interpolation 25a to 25c. How do you know that this means 25a, 25b, 25c? You know by removing the number and then starting with the a go through code points adding one until you reach c. Easy. This will work for all alphabets where that are layed out in alphabetical order in Unicode, and they probably all are. (but thats an assumption on my part :-) But, when you see an interpolation 25 to 25c. How do you know that this means 25, 25a, 25b, 25c? You again remove the number. That gives you the first entry. But what happens then? You count to the end of this interpolation-way (there are 4 nodes in, 1 is the start and 1 is the end, so 2 in between) and substract this number (2) from the last char (c). Following your assumptions from above this should work (when you can add 1 to go to the next char, you can also subtract 1 to go to the previous). This way you'll first get b and then a - those are the chars for your houses. No. The interpolation way has less nodes in it than houses. Thats the whole point of having an interpolation way. Otherwise you'd just use those nodes and tag them with the right house numbers and you are done. Jochen -- Jochen Topf joc...@remote.org http://www.remote.org/jochen/ +49-721-388298 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
2009/10/2 Jochen Topf joc...@remote.org: No. The interpolation way has less nodes in it than houses. Thats the whole point of having an interpolation way. Otherwise you'd just use those nodes and tag them with the right house numbers and you are done. +1. Why not simply use explicit nodes/polygons for every house and you're done. Will be more accurate and faster to just do it instead of discussing possible/real problems with interpolation (which IMHO still is an intermediate step for explicit mapping). cheers, Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On 02/10/2009 12:42, Martin Koppenhoefer wrote: 2009/10/2 Jochen Topf joc...@remote.org: No. The interpolation way has less nodes in it than houses. Thats the whole point of having an interpolation way. Otherwise you'd just use those nodes and tag them with the right house numbers and you are done. +1. Why not simply use explicit nodes/polygons for every house and you're done. Will be more accurate and faster to just do it instead of discussing possible/real problems with interpolation (which IMHO still is an intermediate step for explicit mapping). Yahoo coverage is still very sparse. (And I have also never been able to get it to work since I installed Firefox 3.5). The surveying cost of collecting house numbers is an order of magnitude greater than doing streets/pois and doing every house without satellite coverage is another order of magnitude again as you have to collect locations for every single house. We can't really do polygons without satellite coverage. So while I agree having every house is a desirable solution I think it will be a very long time before this can be a reality other than in satellite coverage areas, and even then it will be a much slower process to relate house numbers to satellite images than to do GPS mapping. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On 28/09/2009, at 2:22 PM, Marcus Wolschon wrote: 25A-25C should work with addr:interpolation=alphabetic . However not all software that supports interpolation at all, supports this interpolation-mode yet. 25-25A would not. I'm not sure you how you can interpolate things like this correctly if you're just using a single interpolation way. For example I've seen both 23-25-25a-27 and 23-25a-25-27, with the 'a' house being whichever one was built second or isn't the primary residence. Whenever I've encountered something like this, I've just broken the way up, so that there is one for ...-23, two plain nodes to 25 and 25a, and another way for 27-. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On Wed, 30 Sep 2009 18:51:35 +1000, James Livingston doc...@mac.com wrote: On 28/09/2009, at 2:22 PM, Marcus Wolschon wrote: 25A-25C should work with addr:interpolation=alphabetic . However not all software that supports interpolation at all, supports this interpolation-mode yet. 25-25A would not. I'm not sure you how you can interpolate things like this correctly if you're just using a single interpolation way. For example I've seen both 23-25-25a-27 and 23-25a-25-27, with the 'a' house being whichever one was built second or isn't the primary residence. Basically...you can't. Whenever I've encountered something like this, I've just broken the way up, so that there is one for ...-23, two plain nodes to 25 and 25a, and another way for 27-. Thats okay. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On 30/09/2009 09:51, James Livingston wrote: On 28/09/2009, at 2:22 PM, Marcus Wolschon wrote: 25A-25C should work with addr:interpolation=alphabetic . However not all software that supports interpolation at all, supports this interpolation-mode yet. 25-25A would not. I'm not sure you how you can interpolate things like this correctly if you're just using a single interpolation way. For example I've seen both 23-25-25a-27 and 23-25a-25-27, with the 'a' house being whichever one was built second or isn't the primary residence. Whenever I've encountered something like this, I've just broken the way up, so that there is one for ...-23, two plain nodes to 25 and 25a, and another way for 27-. I wasn't suggesting that 23 would be included in the interpolation, merely that the first item of an alphabetic interpolation could be no letter. Presumably you could start at B or D, so why not no letter at all, it's not ambiguous. And I think the previous point about other alphabets is a red herring too: basically no letter can precede all other letters in whatever alphabet. If there's ambiguity in the alphabet due to the glyph appearing in more than one, then that's a problem anyway, and would, perhaps need a tag to sequence. (Of course if the way of sequencing houses in another alphabet doesn't follow lexical ordering then all bets are off anyway). It's not a big deal this, it just seems to fit the circumstances more naturally. In the case I was dealing with it would have meant I could have 25-25D 27-27C 29-29C instead of 25 25A-25D 27 27A-27C 29 29A-29C the first non-lettered item belongs naturally in that sequence (especially as on the ground, the each of the 25's, 27's and 29's were physically joined as terraces), and certainly in the UK there's generally no A without a corresponding no-letter in the sequence. OTOH, letters A are pretty unusual, and I am surprised they didn't renumber the street in this instance, as everything nearby had been rebuilt as well and they'd even move a street name from one street to another. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On Tue, Sep 29, 2009 at 8:59 PM, Mike N. nice...@att.net wrote: Since I haven't heard any counter points of view, how do we proceed with this? Based on previous comments, it should not just be Edit the wiki page, or is this change small enough to just update the wiki? If you're referring to defining alphabetic address interpolation (for the latin alphabet), I would say just update the wiki. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On 28 Sep 2009, at 05:22, Marcus Wolschon mar...@wolschon.biz wrote: On Sun, Sep 27, 2009 at 7:33 PM, David Earl da...@frankieandshadow.com wrote: I'm experimenting with adding house numbering for the first time (and using the address interpolation plugin). One common case I came across was 25, 25A, 25B, ... I wonder whether addr:interpolation=alphabetic could include this case 25A-25C should work with addr:interpolation=alphabetic . However not all software that supports interpolation at all, supports this interpolation-mode yet. 25-25A would not. Marcus Yes, that was my reading of the wiki page. I was suggesting that we *should* extend the definition (and the plugin) to include that case. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
What plugin are you talking about? The AdvancedAddressDB of Traveling Salesman? Sorry, the AddrInterpolation plugin in JOSM. http://wiki.openstreetmap.org/wiki/JOSM/Plugins/AddrInterpolation which doesn't allow you to put just a number in the starting # field when numbering scheme is set to alphabetic. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On Sun, Sep 27, 2009 at 06:33:25PM +0100, David Earl wrote: I'm experimenting with adding house numbering for the first time (and using the address interpolation plugin). One common case I came across was 25, 25A, 25B, ... I wonder whether addr:interpolation=alphabetic could include this case (in essence when the first node has no letter, the second would be A, the next B and so on up to the explicit final one, rather than always starting on a letter - though that existing case isn't excluded of course). It doesn't need any change other than in how it is documented. The JOSM plugin could easily support this too. Where it's just 25 and 25A (usually a house that's been built in a gap) it hardly matters, but I found them 3, 4 and 5 in a row like this. Problem is: How do you know where the alphabet starts? If hebrow housenumbers go from 25alef to 25yod, its resonably easy to do this, because the characters in the alphabet will probably be in order in the Unicode list. But if you have the number 25 to 25yod, its harder to find this automatically. Basically you'll need a list of all alphabets with all their characters. I guess we could allow it for the latin alphabet and see where it goes. Jochen -- Jochen Topf joc...@remote.org http://www.remote.org/jochen/ +49-721-388298 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
[OSM-talk] address interpolation
I'm experimenting with adding house numbering for the first time (and using the address interpolation plugin). One common case I came across was 25, 25A, 25B, ... I wonder whether addr:interpolation=alphabetic could include this case (in essence when the first node has no letter, the second would be A, the next B and so on up to the explicit final one, rather than always starting on a letter - though that existing case isn't excluded of course). It doesn't need any change other than in how it is documented. The JOSM plugin could easily support this too. Where it's just 25 and 25A (usually a house that's been built in a gap) it hardly matters, but I found them 3, 4 and 5 in a row like this. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] address interpolation
On Sun, Sep 27, 2009 at 7:33 PM, David Earl da...@frankieandshadow.com wrote: I'm experimenting with adding house numbering for the first time (and using the address interpolation plugin). One common case I came across was 25, 25A, 25B, ... I wonder whether addr:interpolation=alphabetic could include this case 25A-25C should work with addr:interpolation=alphabetic . However not all software that supports interpolation at all, supports this interpolation-mode yet. 25-25A would not. Marcus ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
Thank you Brian for your tips, I edited address with suggestions you made. Can I ask you just to check if I made it ok now, because I will start adding street numbers so I would like to be sure I'm doing ti correctly: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF So what I have done how is wrong and I should put back as it was with tags on nodes? So I should remove: addr:city = Osijek addr:country = Croatia addr:postcode = 31000 addr:street = Starigradska from way and put it back on nodes? According to the wiki as now written (I hate wikis for documentation!) what you have done, and I suggested, is wrong. However there are plenty of cases where people have used it as I suggested because it makes sense and any one implementing reading OSM data for addresses will have to deal with both so in my opinion it makes not one jot of difference. Doing it as I suggested is cleaner (IMO) and avoids duplication but is different to the method originally suggested - but this is OSM so tag it anyway you wish :-) -- Brian ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
So I should remove: addr:city = Osijek addr:country = Croatia addr:postcode = 31000 addr:street = Starigradska from way and put it back on nodes? Imo, yes, you should put all those details onto the objects that carry addr:housenumber (either nodes or building outlines). That's the method intended by the documentation and I don't see a good reason for not sticking to it in this case. Brian Quinion wrote: However there are plenty of cases where people have used it as I suggested because it makes sense It does not make much sense to add information to a temporary construct (interpolation way) that will be replaced with individual tags on each building outline in the long term anyway. and any one implementing reading OSM data for addresses will have to deal with both I think an evaluator can ignore addr:street on interpolation ways - with documentation and tools (such as JOSM presets) supporting consistent tagging you will be able to extract most data this way. Unless, of course, enough people prevent consistent tagging by denying its possibility. Tobias Knerr ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
Imo, yes, you should put all those details onto the objects that carry addr:housenumber (either nodes or building outlines). That's the method intended by the documentation and I don't see a good reason for not sticking to it in this case. inconsistent duplication. I can't image having to convince anyone that this was bad: http://thedailywtf.com/Articles/The-Utlimate-State-Selector.aspx Why is this a special case where duplication is a good thing? It may be that no-one can think of a better way of doing it - but that doesn't make it good. Brian Quinion wrote: However there are plenty of cases where people have used it as I suggested because it makes sense It does not make much sense to add information to a temporary construct (interpolation way) that will be replaced with individual tags on each building outline in the long term anyway. I think in many places this data will not be very temporary. Due to the rapid rate of mapping Germany may well be the exception. and any one implementing reading OSM data for addresses will have to deal with both I think an evaluator can ignore addr:street on interpolation ways - with documentation and tools (such as JOSM presets) supporting consistent tagging you will be able to extract most data this way. Unless, of course, enough people prevent consistent tagging by denying its possibility. Well speaking as an evaluator I can say that simply coping with the possibility of addr:street being on the way rather than the node is very trivial compared with all the other difficulties, in fact it falls out of the code required to cope with the relations anyway. Discarding all data that doesn't perfectly conform to the specification would remove quite a large percentage - this case alone accounts for around 3% of the data. In a way I don't actually care about which is the 'correct' answer, I've written my code to cope with this and a lot of the other edge cases because in practical terms with the current data that is the only choice. It is more that I'm confused by the the apparent assumption that this is the one specification in OSM that will never change - everything else in OSM evolves. -- Brian ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
Brian Quinion wrote: Imo, yes, you should put all those details onto the objects that carry addr:housenumber (either nodes or building outlines). That's the method intended by the documentation and I don't see a good reason for not sticking to it in this case. inconsistent duplication. I'd accept inconsistent duplication as an argument against using addr:street at all and in favor of associatedStreet relations. I'd also accept it for addr:city vs. boundary polygons. But as an argument for moving address details to the interpolation ways? It's at most a factor 2 for duplications, that doesn't really change much. You will want some checks no matter whether it's 1 or 2 addr:street entries in a city. Considering that this question is not decisive for the amount of duplication, other aspects are more relevant here. Namely: - 1 way of doing it is better than 2 ways of doing it - usability (for example: use the same JOSM presets for housenumbers with and without interpolation way attached to them; no partial copy of attributes required when removing an interpolation way) Discarding all data that doesn't perfectly conform to the specification would remove quite a large percentage - this case alone accounts for around 3% of the data. These percentages might decrease if people actually had an incentive to create data conforming to the specification (such as applications only accepting data that does). 3% doesn't sound that impressive, btw. Is that before or after performing interpolation? Tobias Knerr ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
On Wed, Sep 9, 2009 at 10:57 PM, Valent Turkovic valent.turko...@gmail.com wrote: On Tue, Sep 8, 2009 at 2:18 PM, Brian Quinionopenstreet...@brian.quinion.co.uk wrote: Hi, On Tue, Sep 8, 2009 at 1:05 PM, Valent Turkovicvalent.turko...@gmail.com wrote: Hi, I'm using address interpolation for the first time so I would like to ask if somebody can check if I did it ok or if there are some errors: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF They seem OK - and my processing code can interpret them (yeh!) but I'd suggest a couple of changes For http://www.openstreetmap.org/browse/node/489432179 I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska to the way (rather than the individual nodes). And I'd suggest that addr:country = 385 is unlikely to be understood. In general creating a polygon / relation for anything above street level is probably more useful than adding it to individual nodes (or even ways) - so just draw a rough polygon for the city of Osijek and tag that instead. -- Brian Thank you Brian for your tips, I edited address with suggestions you made. Can I ask you just to check if I made it ok now, because I will start adding street numbers so I would like to be sure I'm doing ti correctly: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF Cheers! So what I have done how is wrong and I should put back as it was with tags on nodes? So I should remove: addr:city = Osijek addr:country = Croatia addr:postcode = 31000 addr:street = Starigradska from way and put it back on nodes? -- pratite me na twitteru - www.twitter.com/valentt http://kernelreloaded.blog385.com/ linux, blog, anime, spirituality, windsurf, wireless registered as user #367004 with the Linux Counter, http://counter.li.org. ICQ: 2125241, Skype: valent.turkovic, msn: valent.turko...@hotmail.com ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
On Tue, Sep 08, 2009 at 01:18:33PM +0100, Brian Quinion wrote: On Tue, Sep 8, 2009 at 1:05 PM, Valent Turkovicvalent.turko...@gmail.com wrote: Hi, I'm using address interpolation for the first time so I would like to ask if somebody can check if I did it ok or if there are some errors: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF They seem OK - and my processing code can interpret them (yeh!) but I'd suggest a couple of changes For http://www.openstreetmap.org/browse/node/489432179 I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska to the way (rather than the individual nodes). And I'd suggest that addr:country = 385 is unlikely to be understood. No! Please don't do that. That makes it harder to use. Then there are two possible ways, where data can be. Please use only addr:interpolation on the way and everything else on the nodes. (Of course those tags *can* be on ways, but that has a different meaning: Thats for building outlines that get tagged with an address. Another reason for keeping those things separate.) In general creating a polygon / relation for anything above street level is probably more useful than adding it to individual nodes (or even ways) - so just draw a rough polygon for the city of Osijek and tag that instead. No. Creating polygons and relations willy nilly makes this harder to use. Again, it means there are several places where the software has to look for the data and several places where people have to look for the data. Especially relations are easy to break and easy to overlook, so they should be used rarely. The Karlsruhe schema in its most basic form where all this data is on the node might duplicate lots of data, but it is very robust. Jochen -- Jochen Topf joc...@remote.org http://www.remote.org/jochen/ +49-721-388298 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska to the way (rather than the individual nodes). And I'd suggest that addr:country = 385 is unlikely to be understood. No! Please don't do that. That makes it harder to use. Then there are two possible ways, where data can be. Please use only addr:interpolation on the way and everything else on the nodes. (Of course those tags *can* be on ways, but that has a different meaning: Thats for building outlines that get tagged with an address. Another reason for keeping those things separate.) On the way - addr:interpolation *and* addr:street ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
No! Please don't do that. That makes it harder to use. Then there are two possible ways, where data can be. Please use only addr:interpolation on the way and everything else on the nodes. Which seems to be the opposite of what the section on the Karlsruhe interpolation wiki section says: http://wiki.openstreetmap.org/wiki/Proposed_features/House_numbers/K arlsruhe_Schema#Using_interpolation_to_mark_many_houses_along_a_way (shortened to http://is.gd/34St1 ) That example shows ONLY the house number on the nodes, and everything else on the way used for interpolation. I don't use anything else on the interpolation way however. I put house numbers on end nodes, interpolation on the way linking them, and add that way to an associatedStreet relation for everything else to be worked out. http://wiki.openstreetmap.org/wiki/Proposed_features/House_numbers/K arlsruhe_Schema#Using_Relations_to_associate_house_and_street_.28opt ional.29 (shortened to http://is.gd/34T0q ) Ed ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
http://www.openstreetmap.org/browse/node/489432179 I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska No! Please don't do that. That makes it harder to use. Then there are two possible ways, where data can be. Please use only addr:interpolation on the way and everything else on the nodes. (Of course those tags *can* Have the above details on the nodes makes the data potentially inconsistent because given 2 nodes: node1: addr:street = Starigradska node2: addr:street = SomethingElse way: addr:interpolation = all There is no way to know what street address the interpolated points have. And enough other people are already doing this that assuming that you can ignore tags on the way just doesn't work. Your advise also contradicts the definition on wiki. Putting the tags on the way prevents inconsistency and duplication. In general creating a polygon / relation for anything above street level is probably more useful than adding it to individual nodes (or even ways) - so just draw a rough polygon for the city of Osijek and tag that instead. No. Creating polygons and relations willy nilly makes this harder to use. Again, it means there are several places where the software has to look for the data and several places where people have to look for the data. I disagree with your statement but you have also miss-interpreted what I said. I mean use a polygon / relation to create a polygon for the place (in this case Osijek). The street/house is then known to be within the town because it is inside the town polygon. -- Brian ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
On Wed, Sep 09, 2009 at 02:51:36PM +0100, Ed Loach wrote: No! Please don't do that. That makes it harder to use. Then there are two possible ways, where data can be. Please use only addr:interpolation on the way and everything else on the nodes. Which seems to be the opposite of what the section on the Karlsruhe interpolation wiki section says: http://wiki.openstreetmap.org/wiki/Proposed_features/House_numbers/K arlsruhe_Schema#Using_interpolation_to_mark_many_houses_along_a_way (shortened to http://is.gd/34St1 ) That example shows ONLY the house number on the nodes, and everything else on the way used for interpolation. Somebody must have changes this and I have just changed it back. Jochen -- Jochen Topf joc...@remote.org http://www.remote.org/jochen/ +49-721-388298 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
that you can ignore tags on the way just doesn't work. Your advise Do you have numbers for that? There are, as of last Wednesday: 46899 uses with addr:street in this way I described 209340 uses with addr:street used to link a building outline to a street 2947067 uses with addr:street used to link a node to a street Because of the duplication the 46899 way uses actually relate to 83579 equivalent nodes (for comparison) and while this is definitely a far smaller number than the original usage - it is still large in OSM terms. also contradicts the definition on wiki. Somebody must have changed the Wiki. It used to be different. I have changed it back. I would argue that you have just removed the documentation for how people are using the tag. Putting the tags on the way prevents inconsistency and duplication. Duplication is good. It helps with finding errors. No, duplication is almost always bad (caching may be an exception). Inconsistent data is the enemy of all good database management because you can't tell what it means and if data changes it is easy to miss changing it in multiple places. But this may be a religious war there is no point in having, although I am, of course, right :) I mean use a polygon / relation to create a polygon for the place (in this case Osijek). The street/house is then known to be within the town because it is inside the town polygon. Ok, thats a different issue. If you already have, say, an area with landuse=residential for the town, you could also tag it with this data. But its totally undefined what this is supposed to mean. If people just put those tags anywhere its hard to make sure the right meaning is understood. Depending on whether a way is closed and on other tags this way has, different things could be meant. Say the motorway around London is tagged with addr:postcode, does this mean that everything inside it, has this postcode? Probably not. But what if it is also tagged with a boundary tag? The consensus use of the boundary=administrative relation seems to me to be clearly and (unusually!) consistent. http://wiki.openstreetmap.org/wiki/Key:boundary The meaning of a place polygon is also clearly described: http://wiki.openstreetmap.org/wiki/Key:place -- Brian ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
Hi, Brian Quinion wrote: No, duplication is almost always bad (caching may be an exception). Inconsistent data is the enemy of all good database management *Inconsistent* data is surely not desirable, but *redundant* information may well have its place because it makes it easier to spot errors. Bye Frederik ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
On Tue, Sep 8, 2009 at 2:18 PM, Brian Quinionopenstreet...@brian.quinion.co.uk wrote: Hi, On Tue, Sep 8, 2009 at 1:05 PM, Valent Turkovicvalent.turko...@gmail.com wrote: Hi, I'm using address interpolation for the first time so I would like to ask if somebody can check if I did it ok or if there are some errors: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF They seem OK - and my processing code can interpret them (yeh!) but I'd suggest a couple of changes For http://www.openstreetmap.org/browse/node/489432179 I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska to the way (rather than the individual nodes). And I'd suggest that addr:country = 385 is unlikely to be understood. In general creating a polygon / relation for anything above street level is probably more useful than adding it to individual nodes (or even ways) - so just draw a rough polygon for the city of Osijek and tag that instead. -- Brian Thank you Brian for your tips, I edited address with suggestions you made. Can I ask you just to check if I made it ok now, because I will start adding street numbers so I would like to be sure I'm doing ti correctly: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF Cheers! -- pratite me na twitteru - www.twitter.com/valentt http://kernelreloaded.blog385.com/ linux, blog, anime, spirituality, windsurf, wireless registered as user #367004 with the Linux Counter, http://counter.li.org. ICQ: 2125241, Skype: valent.turkovic, msn: valent.turko...@hotmail.com ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
On Thu, Sep 10, 2009 at 7:28 AM, andrzej zaborowskibalr...@gmail.com wrote: in this case I agree we should stick to the schema the way it was originally defined, good or bad, and I normally only use addr:street on the nodes. +1 Another argument for doing that is that the addr:interpolation way is a temporary placeholder +1 ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
[OSM-talk] Address interpolation
Hi, I'm using address interpolation for the first time so I would like to ask if somebody can check if I did it ok or if there are some errors: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF Thank you in advance! -- pratite me na twitteru - www.twitter.com/valentt http://kernelreloaded.blog385.com/ linux, blog, anime, spirituality, windsurf, wireless registered as user #367004 with the Linux Counter, http://counter.li.org. ICQ: 2125241, Skype: valent.turkovic ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Address interpolation
Hi, On Tue, Sep 8, 2009 at 1:05 PM, Valent Turkovicvalent.turko...@gmail.com wrote: Hi, I'm using address interpolation for the first time so I would like to ask if somebody can check if I did it ok or if there are some errors: http://www.openstreetmap.org/?lat=45.544703lon=18.718653zoom=18layers=B000FTF They seem OK - and my processing code can interpret them (yeh!) but I'd suggest a couple of changes For http://www.openstreetmap.org/browse/node/489432179 I'd suggest moving all the following tags addr:city = Osijek addr:country = 385 addr:postcode = 31000 addr:street = Starigradska to the way (rather than the individual nodes). And I'd suggest that addr:country = 385 is unlikely to be understood. In general creating a polygon / relation for anything above street level is probably more useful than adding it to individual nodes (or even ways) - so just draw a rough polygon for the city of Osijek and tag that instead. -- Brian ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk