Re: [Talk-us] how to handle far away user making global changes
Hi, On 12/14/12 05:54, Peter Dobratz wrote: I guess at this point I would like to pursue reverting these changes, but I'm not sure about what the next step is. I've talked to Shimas and reverted the changes. 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
[Talk-us] Where is this way in the DB ?
This way http://www.openstreetmap.org/browse/way/13694101 is not in ways nor way_nodes tables in a PostGIS replica I have of the OSM planet, yet it appears in the master DB copy. I admit that I cannot understand how this might be, and I'm wondering if anyone has more insight into this situation. Where's the data ? Thanks, -- John Novak 585-OLD-TOPOS (585-653-8676) http://www.linkedin.com/in/johnanovak/ OSM ID:oldtopos OSM Heat Map: http://yosmhm.neis-one.org/?oldtopos OSM Edit Stats:http://hdyc.neis-one.org/?oldtopos ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Where is this way in the DB ?
Are you running a snapshot schema, imported with osmosis? If so then you just discovered the same thing I did a couple of months ago. Osmosis silently drops ways with less than 2 nodes during import. (yes, ways with zero nodes exist too) This is because they create an invalid linestring which can cause problems with geographic queries. There is one plot twist. While consuming replication diffs, osmosis does *not* drop these ways. So you probably do have zero and single node ways in your database, but only ones created after you started applying minutely/hourly/daily diffs. I have written a patch for osmosis that makes this behavior explicit and optional with a --kepInvalidWays=yes/no option. It also allows the same option to be applied to diff consumption so that the two are consistent. It has yet to be merged as there was some discussion about it on the osmosis-dev mailing list. It is on my github fork though: https://github.com/ToeBee/osmosis Unfortunately P2 has a bug that creates these ways rather often. OSM inspector has a layer that displays them: http://tools.geofabrik.de/osmi/?view=geometryoverlays=single_node_in_way IMO the editing API really should reject ways with zero or one node in them as invalid. But there is no way that change will happen until P2 is fixed... Toby On Fri, Dec 28, 2012 at 1:47 PM, the Old Topo Depot oldto...@novacell.com wrote: This way http://www.openstreetmap.org/browse/way/13694101 is not in ways nor way_nodes tables in a PostGIS replica I have of the OSM planet, yet it appears in the master DB copy. I admit that I cannot understand how this might be, and I'm wondering if anyone has more insight into this situation. Where's the data ? Thanks, -- John Novak 585-OLD-TOPOS (585-653-8676) http://www.linkedin.com/in/johnanovak/ OSM ID:oldtopos OSM Heat Map: http://yosmhm.neis-one.org/?oldtopos OSM Edit Stats:http://hdyc.neis-one.org/?oldtopos ___ 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] Where is this way in the DB ?
Yes, yes I am. This explains why, when I checked timestamps on the zero/one node ways, they were all after the timestamp on the planet file I pulled to create the initial planet DB. It also explains the difference between the Geofabrik errors and the query results returned from the internal planet DB. https://trac.openstreetmap.org/ticket/4378 documents the Potlatch issue, which will be challenging to fully resolve due to the apparent number of variants in the wild. I have no .as foo or I'd try to fix it myself. Yes, an API fix to reject zero/one node ways might be best, as it's gonna get tedious to keep clearing these errors. Dare I suggest a fixbot ;-) Best, On Fri, Dec 28, 2012 at 12:11 PM, Toby Murray toby.mur...@gmail.com wrote: Are you running a snapshot schema, imported with osmosis? If so then you just discovered the same thing I did a couple of months ago. Osmosis silently drops ways with less than 2 nodes during import. (yes, ways with zero nodes exist too) This is because they create an invalid linestring which can cause problems with geographic queries. There is one plot twist. While consuming replication diffs, osmosis does *not* drop these ways. So you probably do have zero and single node ways in your database, but only ones created after you started applying minutely/hourly/daily diffs. I have written a patch for osmosis that makes this behavior explicit and optional with a --kepInvalidWays=yes/no option. It also allows the same option to be applied to diff consumption so that the two are consistent. It has yet to be merged as there was some discussion about it on the osmosis-dev mailing list. It is on my github fork though: https://github.com/ToeBee/osmosis Unfortunately P2 has a bug that creates these ways rather often. OSM inspector has a layer that displays them: http://tools.geofabrik.de/osmi/?view=geometryoverlays=single_node_in_way IMO the editing API really should reject ways with zero or one node in them as invalid. But there is no way that change will happen until P2 is fixed... Toby On Fri, Dec 28, 2012 at 1:47 PM, the Old Topo Depot oldto...@novacell.com wrote: This way http://www.openstreetmap.org/browse/way/13694101 is not in ways nor way_nodes tables in a PostGIS replica I have of the OSM planet, yet it appears in the master DB copy. I admit that I cannot understand how this might be, and I'm wondering if anyone has more insight into this situation. Where's the data ? Thanks, -- John Novak 585-OLD-TOPOS (585-653-8676) http://www.linkedin.com/in/johnanovak/ OSM ID:oldtopos OSM Heat Map: http://yosmhm.neis-one.org/?oldtopos OSM Edit Stats:http://hdyc.neis-one.org/?oldtopos ___ 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 -- John Novak 585-OLD-TOPOS (585-653-8676) http://www.linkedin.com/in/johnanovak/ OSM ID:oldtopos OSM Heat Map: http://yosmhm.neis-one.org/?oldtopos OSM Edit Stats:http://hdyc.neis-one.org/?oldtopos ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
On Fri, Dec 28, 2012 at 4:16 PM, Jason Remillard remillard.ja...@gmail.com wrote: The open space layer from MassGIS was imported several years ago. This has encouraged people to map out many of the hiking trails. How do you make the connection from The MassGIS open space layer was imported to this has encouraged people to map out many the hiking trails? You're asserting a causal relationship where I see two unrelated events. Surprisingly, I think that OSM is currently the best/most complete map of hiking trails in Mass. Why is that surprising? OSM is the most compete biking map of the UK, and Germany. Outdoor activities and OSM have a very close relationship. In fact many of the mappers in Mass came to OSM from the local trail committee's (myself included). So reality is that we do have some parcel data data in OSM and its inclusion has been a net positive. Can you explain how the parcel data is a net positive? Also, can you explain why the import there is better than what was done in most of the rest of the world, where mappers simply went out with a GPS and mapped out the hiking trails themselves? - Serge ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
* Jason Remillard remillard.ja...@gmail.com [2012-12-28 16:16 -0500]: So the question is, what should the exact criteria be for including an open space parcel in OSM. Consider some of the various types of property. I've used parcel data as a layer in JOSM to trace from. It lets me be a little more accourate about some area boundaries than I could from just aerial imagery (and walking a GPS along, say, the border between a golf course and a residential area with private houses is a little out of the question). I'd be resistant to the idea of bulk import (pretty much anything beyond pulling individually-checked polygons into OSM) because I've seen a lot of places where a naive import of the parcel data available would have made for wrong or at least weird OSM data. I've seen a number of places where a single entity acquired its land over time, so the parcel records show multiple parcels that should be a single OSM entity. Similarly, I've seen a lot of places where a public road cuts through a single entity's land (golf courses especially, but also parks and residential areas). I feel it's more correct to make a single polygon that crosses the road, but parcel data would usually have the road splitting the area. I've also seen a few places where parcels were too broad, where a single parcel needs to be divided into several different OSM landuses. This is just my experience with the handful of counties in Maryland that have parcel data available under an OSM-friendly license. Maybe other jurisdictions have data that would provide a more one-to-one correspondence with OSm features. Even in those cases, an importer would need to make sure that the import fits topologically into OSM, interacting properly with existing data. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Parcel data in and of itself are not inherently bad to have in OSM as long as they are filtered and modified before adding. For instance an open space parcel probably isn't that useful because it is not represented in OSM. It could be broken up into meadow, wood, scrub, forest, etc. Other parcel data that don't translate well include things like flood planes or highway zones. Even some forest parcels may not always translate into landuse=forest. Within cities, you have parcels that are subdivided into areas like medium family residential, multi family residential, non retail commercial, mixed use, etc. Parcel data tends to be too vague or tends to overlap other features. A better way to add it is to filter out each individual feature first, verify them and then upload them individually. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Hi Serge, To answer your questions, consider the following - Most of the hiking trails in MA were put in over the imported open space layer. - Unlike other countries, It is unacceptably risky to go on a hike on some random trail that might be on private property. You are likely to find yourself in an unpleasant confrontation with the property owner. Believe me, I love talking about imports as much as anybody :-) but, I was hoping to talk more about the 2nd half of the email. When I started with OSM, the import has long since been done and I was hoping to have a discussion about how to deal with the existing data. I have been working on it, but I am not sure if I am doing the best thing. Thanks Jason. On Fri, Dec 28, 2012 at 4:33 PM, Serge Wroclawski emac...@gmail.com wrote: On Fri, Dec 28, 2012 at 4:16 PM, Jason Remillard remillard.ja...@gmail.com wrote: The open space layer from MassGIS was imported several years ago. This has encouraged people to map out many of the hiking trails. How do you make the connection from The MassGIS open space layer was imported to this has encouraged people to map out many the hiking trails? You're asserting a causal relationship where I see two unrelated events. Surprisingly, I think that OSM is currently the best/most complete map of hiking trails in Mass. Why is that surprising? OSM is the most compete biking map of the UK, and Germany. Outdoor activities and OSM have a very close relationship. In fact many of the mappers in Mass came to OSM from the local trail committee's (myself included). So reality is that we do have some parcel data data in OSM and its inclusion has been a net positive. Can you explain how the parcel data is a net positive? Also, can you explain why the import there is better than what was done in most of the rest of the world, where mappers simply went out with a GPS and mapped out the hiking trails themselves? - Serge ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
On 12/28/2012 4:47 PM, Phil! Gold wrote: * Jason Remillard remillard.ja...@gmail.com [2012-12-28 16:16 -0500]: So the question is, what should the exact criteria be for including an open space parcel in OSM. Consider some of the various types of property. I've used parcel data as a layer in JOSM to trace from. It lets me be a little more accourate about some area boundaries than I could from just aerial imagery (and walking a GPS along, say, the border between a golf course and a residential area with private houses is a little out of the question). I'd be resistant to the idea of bulk import (pretty much anything beyond pulling individually-checked polygons into OSM) because I've seen a lot of places where a naive import of the parcel data available would have made for wrong or at least weird OSM data. I've seen a number of places where a single entity acquired its land over time, so the parcel records show multiple parcels that should be a single OSM entity. Similarly, I've seen a lot of places where a public road cuts through a single entity's land (golf courses especially, but also parks and residential areas). I feel it's more correct to make a single polygon that crosses the road, but parcel data would usually have the road splitting the area. I've also seen a few places where parcels were too broad, where a single parcel needs to be divided into several different OSM landuses. This is just my experience with the handful of counties in Maryland that have parcel data available under an OSM-friendly license. Maybe other jurisdictions have data that would provide a more one-to-one correspondence with OSm features. Even in those cases, an importer would need to make sure that the import fits topologically into OSM, interacting properly with existing data. +1 on all points. I've seen the same things in FL and use parcels as a backdrop in JOSM to help guide hand digitizing boundaries for things like parks, golf, schools, hospitals, retail, residential areas, etc. And as Phil said, sometimes it doesn't make sense to follow the parcel lines exactly, such as if the parcel boundary extends into a road and it makes more sense to draw the boundary where the park area appears to end some distance from the road. Brian ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
On Fri, Dec 28, 2012 at 1:33 PM, Serge Wroclawski emac...@gmail.com wrote: The open space layer from MassGIS was imported several years ago. This has encouraged people to map out many of the hiking trails. How do you make the connection from The MassGIS open space layer was imported to this has encouraged people to map out many the hiking trails? You're asserting a causal relationship where I see two unrelated events. Maybe it's only evident when you have local knowledge? ; ) Seriously, I could imagine a situation where handing people a map with parks outlined and asking them to fill in the trails with GPX tracks might be a bit less daunting than giving them a blank map and asking them to figure out park boundaries (usually something that you would defer to an expert to share) as well as the trails. But, that's just my guess - I'm curious what Jason's perspective might be. -- Jeff Meyer Global World History Atlas www.gwhat.org j...@gwhat.org 206-676-2347 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Hi, On 28.12.2012 22:16, Jason Remillard wrote: So the question is, what should the exact criteria be for including an open space parcel in OSM. Consider some of the various types of property. I'd say anything that is observable on the ground is fine to map. So if there's a fence around a parcel or some other demarcation then you can add it, but if it's just a line in some government database then don't - because it wouldn't make sense for mappers to edit it anyway, and stuff that is only used for reference when mapping should be a background layer in your editor and not part of the OSM database. 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] parcel data in OSM
Hi Everybody, Frederik I have been doing the background layer/tracing over technique. So, Frederik's, says no to all of these parcels types. Not much gray area in Frederik criteria. - True conservation land, land that is owned by a private non-profit or owned by the town that is supposed to be never developed, the public is allowed to use it for light recreation activities, and that's it. - Town land that is open to the public, but is not developed. Watersheds, parks, undeveloped tracks etc. - Playgrounds - Public Schools - Private land that is open to the public as long as people stay on the marked trails. - Private land that has development restrictions, but is not open to the public. - In between, places like the New England Forestry Foundation, that harvest tree's, so the land is in fact a forest, but encourages the public use the land and who's mission is conservation. Without having a parcel layer, in the openstreetmap.org, the map would be less useful to people who are trying to figure out where to take a hike. Thanks Jason. On Fri, Dec 28, 2012 at 6:11 PM, Frederik Ramm frede...@remote.org wrote: Hi, On 28.12.2012 22:16, Jason Remillard wrote: So the question is, what should the exact criteria be for including an open space parcel in OSM. Consider some of the various types of property. I'd say anything that is observable on the ground is fine to map. So if there's a fence around a parcel or some other demarcation then you can add it, but if it's just a line in some government database then don't - because it wouldn't make sense for mappers to edit it anyway, and stuff that is only used for reference when mapping should be a background layer in your editor and not part of the OSM database. 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 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Serge Wroclawski emac...@gmail.com writes: On Fri, Dec 28, 2012 at 5:25 PM, Brian May b...@mapwise.com wrote: And as Phil said, sometimes it doesn't make sense to follow the parcel lines exactly, such as if the parcel boundary extends into a road and it makes more sense to draw the boundary where the park area appears to end some distance from the road. That's a good point in the general case. In Mass, parcels stop outside the road and the roads are their own parcels. So for a conservation area, the legally protected area is exactly the parcel. So it sounds like what you folks would want is for data where it's available, some sort of tracing background layer, or else a per object import where you could load the data in your editor of choice and manually select how they go in? That seems both doable and non-controversial. I think what Jason was suggesting was looking at a small number of individual parcel polygons, and then, based on local knowledge, deciding to add them to OSM with appropriate tags. As an example, I've done this for two conservation properties in my town. I knew which parcels they were, in one case because I had already hiked it and mapped the tails, and in another because I belong to the group that bought them and had seen maps (from massgis data) showing which parcels were being acquired. So I found them from the assessor's parcel database (massgis, usual PD-attribution-requested), and copied them, adding tags. This is totally different from adding all parcels. With respect to Jason's suggestion that there's a causal link between the open-space layer being imported and people mapping trails, I think it's as strong as the theory that says imports discourage new contributors :-) By that I mean that I don't really believe either theory, but it's interesting fodder for discussions. In my case, his suggestion of a link rings true for me; I became interested in OSM when a friend pointed out that there was already road data and the open space layer in Mass. But to answer the real underlying question, one has to analyze the ensemble of all people who might map, and that's a different and much harder question. My own take on which parcels to import are to do it by hand only, but if one is adding a landuse tag for an area that logically should line up with a parcel boundary, it makes sense to use the best representation of the parcel. So that includes formal conservation areas, but also forestry plots and anything else where a landuse tag boundary is logically associated with parcel boundaries. It's really not possible to find out the boundaries by seeing anything in the woods, and hiring a surveyor to stake them out and then measure the corners is expensive (and probably not doable, on other's land). pgp3M8mloQ1CR.pgp Description: PGP signature ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Nathan Mixter nmix...@gmail.com writes: For instance an open space parcel probably isn't that useful because it is not represented in OSM. It could be broken up into meadow, wood, scrub, forest, etc. Jason and I are using 'open space' to mean land that is protected from development with some legal mechanism; this is orthogonal to landcover. Other parcel data that don't translate well include things like flood planes or highway zones. Even some forest parcels may not always translate into landuse=forest. Agreed; per-case mapper judgement is needed. Within cities, you have parcels that are subdivided into areas like medium family residential, multi family residential, non retail That sounds like zoning rather than parcel. commercial, mixed use, etc. Parcel data tends to be too vague or tends to overlap other features. A better way to add it is to filter out each individual feature first, verify them and then upload them individually. Here's a specific example of what I'm talking about:b http://osm.org/go/ZfIZMM6m The area labeled Leggett is a single parcel (lot), owned by my local land trust. The protected area is exactly what the trust owns, and the massgis lot data is the best available set of coordinates, absent hiring surveyors (in this case, in my judgement, having walked the land and knowing most of the board of the land trust). pgpdTcHwZx1WX.pgp Description: PGP signature ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
On Fri, Dec 28, 2012 at 9:22 PM, Russ Nelson nel...@crynwr.com wrote: Frederik Ramm writes: add it, but if it's just a line in some government database then don't - Here in the US where you aren't allowed to trespass on private property except on certain conditions, these line[s] in some government database MATTER to mappers and to map users. It may be different where you live. That doesn't mean you should advocate for us to map badly. Oh come on, this has nothing to do with trespassing. Frederik's point is that you should only map things that other mappers can verify or improve on. Since you can't verify borders and boundaries or otherwise make them any better than the government data after they're imported, they don't belong in OSM. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Ian Dees writes: Frederik's point is that you should only map things that other mappers can verify or improve on. Since you can't verify borders and boundaries or otherwise make them any better than the government data after they're imported, they don't belong in OSM. Anybody can verify that the data is accurate by checking it against the original. Next objection? No, seriously, I *do* use the governmental data that Frederick and apparently Ian is objecting to, and I use it by loading it out of OSM. This is not a new discussion, and we didn't come to a conclusion satisfactory to all parties the last N times we had it. -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
On Fri, Dec 28, 2012 at 10:22 PM, Russ Nelson nel...@crynwr.com wrote: Ian Dees writes: Frederik's point is that you should only map things that other mappers can verify or improve on. Since you can't verify borders and boundaries or otherwise make them any better than the government data after they're imported, they don't belong in OSM. Anybody can verify that the data is accurate by checking it against the original. Next objection? The moment it makes its way in to OSM it becomes incorrect. There is *absolutely* no way to improve the data once it's in OSM, so it should not be in OSM. Period. No, seriously, I *do* use the governmental data that Frederick and apparently Ian is objecting to, and I use it by loading it out of OSM. Excellent. Feel free to convert the data into whatever format you desire and use it locally. But it doesn't belong in OSM. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
Ian Dees writes: The moment it makes its way in to OSM it becomes incorrect. There is *absolutely* no way to improve the data once it's in OSM, so it should not be in OSM. Period. That's a great theory, but I don't think many people subscribe to it. Of course anybody can improve on imported data by tagging it, even if its location is already perfect in every way, which, being a database created by humans, is not even remotely likely. There is no point in having this discussion again unless you're going to bring up something new. So far, not. OSM is a big tent with room for lots of data and lots of opinions. You're welcome to express yours, but you're not welcome to claim it is or should be the ruling opinion. -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] parcel data in OSM
So it sounds like what you folks would want is for data where it's available, some sort of tracing background layer, or else a per object import where you could load the data in your editor of choice and manually select how they go in? Yes ... especially for 'large' imports. Looking at it from a content management perspective, for large imports, enabling some sort of reference -- tracing -- staging -- review -- publish capability would be desirable, since unwinding anything after the fact seems to be very difficult. It would make it much less scary folks. Depending on the license status, potentially a dataset could be locked (or available as tiles) for trace only if adequate permissions were not in place, or pending. That seems both doable and non-controversial. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us