[Talk-us] city polygons too large for potlatch to handle?
Hi - This might not be the right group to direct this technical question - but I'll put it out there anyhow. I noticed a little while ago that city polygons where added to the OSM database (at least in the SF Bay Area) - and that's a good thing. There is a city boundary that runs along a major road between San Jose and Campbell that I meant to clean up, by getting it to run along the median of the street. I was also going to redigitize this road as the dual carriageway that it is. Here's a junction between Bascom Ave, California Highway 85, and some of the city polygon boundaries: http://www.openstreetmap.com/?lat=37.254578lon=-121.951574zoom=18layers=B000FTF Every time I try to edit (or even select) the city polygon - to delete unnecesary points, or to get it to run nicely up the middle of the road - potlatch gets stuck in a loop. It eventually shows me a warning: A script in this movie is causing Adobe Flash Player 10 to run slowly. If it continues to run, your computer may become unresponsive. Do you wish to abort the script? After which, it's impossible to complete the edit. This problem is making it difficult to clean up this street. Are the developers aware of this problem? Is there anything that can be done? I was hoping that the new version of Potlatch would correct the problem, but it's not the case. -Alan___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] city polygons too large for potlatch to handle?
On Tue, Apr 28, 2009 at 2:02 AM, Alan Brown adbrown1...@yahoo.com wrote: This might not be the right group to direct this technical question - but I'll put it out there anyhow. I noticed a little while ago that city polygons where added to the OSM database (at least in the SF Bay Area) - and that's a good thing. There is a city boundary that runs along a major road between San Jose and Campbell that I meant to clean up, by getting it to run along the median of the street. I was also going to redigitize this road as the dual carriageway that it is. Here's a junction between Bascom Ave, California Highway 85, and some of the city polygon boundaries: http://www.openstreetmap.com/?lat=37.254578lon=-121.951574zoom=18layers=B000FTF Every time I try to edit (or even select) the city polygon - to delete unnecesary points, or to get it to run nicely up the middle of the road - potlatch gets stuck in a loop. It eventually shows me a warning: A script in this movie is causing Adobe Flash Player 10 to run slowly. If it continues to run, your computer may become unresponsive. Do you wish to abort the script? After which, it's impossible to complete the edit. This problem is making it difficult to clean up this street. Are the developers aware of this problem? Is there anything that can be done? I was hoping that the new version of Potlatch would correct the problem, but it's not the case. Ick. It's possible the OSM API 0.6 upgrade brought in polygons ways that are now too big to edit because they're over the 0.6 limit of 2000 nodes per way--it is not at all clear what the migration for existing polygons/ways was. You may need to use JOSM to do at least a basic edit on these polygons then upload and continue work in Potlatch. (Alternative theory: Potlatch has also just been generally flaky for me post-upgrade, so it could just be Potlatch-flakiness.) Worst comes to worst I can try to blow away the California upload and upload a 0.6-friendly version. I have a very nice, mostly-way-based conversion setup respecting the 0.6 API limits and using complex multipolygon relations that I implemented starting with the Idaho (complete) and Illinois (uploading now) boundaries; ideally I'd like to go back and redo the boundaries in the A-H states if I can think of a sensible way to do it. (Texas I will have to blow away and redo anyway, which I'm not looking forward to.) Chris ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] city polygons too large for potlatch to handle?
Alan Brown wrote: Every time I try to edit (or even select) the city polygon - to delete unnecesary points, or to get it to run nicely up the middle of the road - potlatch gets stuck in a loop. It eventually shows me a warning: A script in this movie is causing Adobe Flash Player 10 to run slowly. If it continues to run, your computer may become unresponsive. Do you wish to abort the script? After which, it's impossible to complete the edit. Yikes. That way is 8756 nodes long. Potlatch isn't ever going to be able to edit a way of that size, I'm afraid. You either need to use an offline editor like JOSM, which doesn't have the restriction of running in a resource-limited browser plugin, or to get someone (preferably the person who imported it ;) ) to write a script to split it into ways of 2000 nodes or less, which is now the limit. (IMO 1000 is a best-practice maximum anyway.) cheers Richard (Potlatch developer) -- View this message in context: http://www.nabble.com/city-polygons-too-large-for-potlatch-to-handle--tp23271316p23274750.html Sent from the Talk-US mailing list archive at Nabble.com. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] city polygons too large for potlatch to handle?
Rather than blowing it away, perhaps there should be a tool to automatically cut and replace large polygons with multiple smaller polygons? Perhaps someone could write a routine using GPC: http://www.cs.man.ac.uk/~toby/alan/software/ . Years ago, when I worked at Etak, the internal format (mapengine) had a limit of 255 points per polygon; all software interfacing with it had to maintain this limit. The limit doesn't have to be that low, but perhaps it should be low enough for Potlatch to handle anything. Of course, the downside is that you can't assume that a polygon's border will the border of the administrative unit or physical feature represented. (How is this handled for Oceans?) The boundary polylines and polygons have to be distinct. -Alan From: Chris Lawrence lordsu...@gmail.com To: OpenStreetMap U.S. talk-us@openstreetmap.org Sent: Tuesday, April 28, 2009 12:17:58 AM Subject: Re: [Talk-us] city polygons too large for potlatch to handle? On Tue, Apr 28, 2009 at 2:02 AM, Alan Brown adbrown1...@yahoo.com wrote: This might not be the right group to direct this technical question - but I'll put it out there anyhow. I noticed a little while ago that city polygons where added to the OSM database (at least in the SF Bay Area) - and that's a good thing. There is a city boundary that runs along a major road between San Jose and Campbell that I meant to clean up, by getting it to run along the median of the street. I was also going to redigitize this road as the dual carriageway that it is. Here's a junction between Bascom Ave, California Highway 85, and some of the city polygon boundaries: http://www.openstreetmap.com/?lat=37.254578lon=-121.951574zoom=18layers=B000FTF Every time I try to edit (or even select) the city polygon - to delete unnecesary points, or to get it to run nicely up the middle of the road - potlatch gets stuck in a loop. It eventually shows me a warning: A script in this movie is causing Adobe Flash Player 10 to run slowly. If it continues to run, your computer may become unresponsive. Do you wish to abort the script? After which, it's impossible to complete the edit. This problem is making it difficult to clean up this street. Are the developers aware of this problem? Is there anything that can be done? I was hoping that the new version of Potlatch would correct the problem, but it's not the case. Ick. It's possible the OSM API 0.6 upgrade brought in polygons ways that are now too big to edit because they're over the 0.6 limit of 2000 nodes per way--it is not at all clear what the migration for existing polygons/ways was. You may need to use JOSM to do at least a basic edit on these polygons then upload and continue work in Potlatch. (Alternative theory: Potlatch has also just been generally flaky for me post-upgrade, so it could just be Potlatch-flakiness.) Worst comes to worst I can try to blow away the California upload and upload a 0.6-friendly version. I have a very nice, mostly-way-based conversion setup respecting the 0.6 API limits and using complex multipolygon relations that I implemented starting with the Idaho (complete) and Illinois (uploading now) boundaries; ideally I'd like to go back and redo the boundaries in the A-H states if I can think of a sensible way to do it. (Texas I will have to blow away and redo anyway, which I'm not looking forward to.) Chris ___ 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
[Talk-us] Mysterious 403s on TIGER place bulk uploads
Is there a good reason why my TIGER place bulk uploads keep getting blocked with 403s? I'll gladly add further backoffs and more reuse of existing changesets (currently I've set the uploader to chunks of 250 elements and changesets of 20 uploads, essentially limiting each changeset to 5000 elements, with a 15 second delay before retrying a 500 error and bailing after 3 consecutive 500s or any other error), or even add a delay between each chunk if the server needs to breathe/reduce contention, but it'd be nice if someone *told* me what the problem was so I could fix it. (BTW, my hacked-up uploader is at http://www.lordsutch.com/osm/bulk_upload.py if anyone has any questions about the code - it's based on the existing Python bulk uploader but with some tweaks to make it more reliable.) Chris ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Mysterious 403s on TIGER place bulk uploads
On Tue, Apr 28, 2009 at 3:21 PM, Chris Lawrence lordsu...@gmail.com wrote: Is there a good reason why my TIGER place bulk uploads keep getting blocked with 403s? I'll gladly add further backoffs and more reuse of existing changesets (currently I've set the uploader to chunks of 250 elements and changesets of 20 uploads, essentially limiting each changeset to 5000 elements, with a 15 second delay before retrying a 500 error and bailing after 3 consecutive 500s or any other error), or even add a delay between each chunk if the server needs to breathe/reduce contention, but it'd be nice if someone *told* me what the problem was so I could fix it. I'm having problems with my NHD uploads, too. I started with 50k elements in each changeset (like the capabilities API says) and have backed all the way down to 2. The problem is that the server accepted one of the 20k-element changesets, so I have to keep using those so the IDs don't get screwed up. Perhaps it's the clients fault, but I don't see any description of why the upload failed returned back (using JOSM in this case). It just keeps retrying. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
[Talk-us] Massachusetts topo map
Hi OSMappers! I've been working on a topographic map based on OSM data, somewhat similar in style to the National Geographic topo maps. It's very much work in progress, but I thought someone here might be interested. Feedback is welcome! It's (currently) limited to the state of Massachusetts - partly because that's where I live and partly because high-resolution contours and elevation data (significantly higher resolution than e.g. SRTM) is freely available through MassGIS. If similar data is available for other regions, the same type of map could most likely be generated there. When I have time I'll fix some of the issues with the map and include the steps (and scripts) used to generate it. Meanwhile, have a look: http://toposm.com/ - Lars PS. I'm looking for hosting. It's a fair amount of data (several GB), but if anyone has the space and bandwidth... -- Lars Ahlzen l...@ahlzen.com ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Massachusetts topo map
I've been working on a topographic map based on OSM data, It's (currently) limited to the state of Massachusetts - fun! Very pretty. will need permalinks, I know a page that should link your render of Fourth Cliff. Day Blvd Beach in South Boston is rendering as a Swamp, as is Malibu Beach? nice look over all, and in detail too. i like it Bill in Boston n1...@arrl.net bill.n1...@gmail.com ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] Massachusetts topo map
Bill Ricker wrote: will need permalinks, I know a page that should link your render of Fourth Cliff. Done. Day Blvd Beach in South Boston is rendering as a Swamp, as is Malibu Beach? I guess so. These areas are classified as FLATS (i.e. tidal flats) in the MassGIS hydrography dataset, where that data comes from. I don't know if the swamp pattern makes sense, but in most cases those are marshes or muddy areas. There seems to be few actual beaches in Mass in the OSM database. nice look over all, and in detail too. i like it Thanks! - Lars -- Lars Ahlzen l...@ahlzen.com ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us