Re: [OSM-dev] Nominatim results when limit=1
I have a service setup using the OSM Nominatim service and I have a query on number of results returned with the limit= parameter. I have it set as 1 to only get the top result BUT when I search for San Francisco or Boston (where countrycodes=us) I surprisingly got no results? I change this to limit=2 and limit=3 and see that I get 1 and 2 results respectively - which seems offset to me... I see the problem - the new code from removing duplicate places occurs too late in the process after the limit has already been applied. Should be an easy fix, I'll get it done and released. Thanks, -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Nominatim data and code updates
tl;dr: nominatim is updated (data and code), add wikipedia tags and label/admin_centre relation member to improve search quality. As some of you may have noticed Sarah (lonvia) has been busy updating the openstreetmap nominatim instance. It should now be fully up to date and back running live updates, so, if anyone is still finding data that they have added is not included please let us know by email on the geocoding mailing list, via trac (select nominatim component) or on either the #osm or #osm-nominatim irc channels. In the process of updating the data we also took the opportunity to release some code changes. Probably the most visible changes in this update are modification in how address hierarchy is calculated, a new technique for calculating the importance using wikipedia articles and a new system for deduplicating place and admin areas. As a result of these changes nominatim now supports a few extra tags that were previously ignored: 'wikipedia' tag [1] and its variants. Adding this allows nominatim to have a far better value for how important a place is which helps with the ordering, for instance a search for statue of liberty now consistently returns the correct one [2] first rather than the on on a traffic island in the UK [3]. If you find a place where the ordering of results is still bad consider adding suitable wikipedia links to the osm elements to help improve scoring. The new version of nominatim also supports both 'label' and 'admin_centre' relation members for boundary relations [4] which allows it to reduce the amount of duplicated data returned and produce a more consistent result set. The 'label' member will be merged (name and tags), relation tags win if there is a conflict. 'admin_centre' member will be merged only if the names and 'rank' (effectively admin_level / place=*) match, but other than that same rules. In both cases the node will also be added as the centre point of the polygon - i.e. the location that the map will centre on and this is returned in preference to the geometric centre of the polygon. If no 'admin_centre' and 'label' members are present the code will try to guess by looking for a node at the right admin level, name and approximate location - obviously explicit tagging is far better and more accurate. Thanks to all those who help with this update, in particular Sarah Hoffmann who did by far the majority of the work! -- Brian [1] http://wiki.openstreetmap.org/wiki/Key:wikipedia [2] http://www.openstreetmap.org/browse/way/32965412 [3] http://www.openstreetmap.org/browse/node/355219404 [4] http://wiki.openstreetmap.org/wiki/Relation:boundary#Relation_members ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] e-control.at (Spritpreisdatenbank) - Nominatim
Andreas, They have indeed been blocked - something I never thought we'd have to do with MQ sponsoring the hardware. They were putting through around 250 search requests per SECOND using what looks like a broken autocomplete implementation. I had to take emergency action to restore the service. I'd be grateful if you could pass this on to your contact at e-control.at - once they have fixed their problems I'll happily remove the block. Cheers, -- Brian On 22 August 2011 13:01, Andreas Labres a...@lab.at wrote: Hi! I was contacted by some guys from e-control.at (they develop some fuel cost database) that they would use Nominatim and their app would be blocked by nominatim. Is this the case? Did someone there contact any sysadmin already? /al ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Nominatim updates paused
Hi, The live install of nominatim has run out of integers on it's primary key and the database is being updated to use bigint. Sorry for the lack of notice - this rather caught me by surprise! Until this is completed there will be no updates. Because of the size of the database and the software changes needed to support this it will probably take most of next week to complete - as such the ETA for completing this work is 13th of June. Apart from updates the service should operate as normal. Anyone using the MapQuest NPI (pre-indexed files) will need to run an update script to update the fields in their local database - I'll post this as soon as I've finished testing it locally. Cheers, -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Daily diff fails to import
Adrian, I had this occur once with a schema that I had updated between versions of postgis. I never did entirely get to the bottom of it - I ended up adding ::text in a similar way to what Sandro suggested. i.e.: existing.geometry::text != NEW.geometry::text It was a crude hack (and probably hurt performance) and didn't seem to be needed the result of the time so I never committed it. I think there were 3 places were it was needed all in the place_update function. -- Brian On 26 May 2011 14:34, Adrian Chapela achapela.rexist...@gmail.com wrote: I am trying to import a daily diff and is impossible. I have tried with util.update.php and the result is the same. Now I am trying to the next command. Below you coud see the output. osm2pgsql -as -O gazetteer -d gazetteer -U postgres -W 20110405-20110406.osc.gz osm2pgsql SVN version 0.70.5 Password: Using projection SRS 900913 (Spherical Mercator) Mid: pgsql, scale=100, cache=800MB, maxblocks=102401*8192 Setting up table: planet_osm_nodes Setting up table: planet_osm_ways Setting up table: planet_osm_rels Reading in file: 20110405-20110406.osc.gz COPY_END for place failed: ERROR: operator is not unique: geometry geometry LINE 5: OR (existing.geometry != NEW.geometry AND ST_Distance(S... ^ HINT: Could not choose a best candidate operator. You might need to add explicit type casts. QUERY: SELECT existing.osm_type IS NULL OR existingplacex.osm_type IS NULL OR coalesce(existing.admin_level, 100) != coalesce(NEW.admin_level, 100) -- OR coalesce(existing.country_code, '') != coalesce(NEW.country_code, '') OR (existing.geometry != NEW.geometry AND ST_Distance(ST_Centroid(existing.geometry),ST_Centroid(NEW.geometry)) 0.01 AND NOT (ST_GeometryType(existing.geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon'))) CONTEXT: PL/pgSQL function place_insert line 52 at IF COPY place, line 1: \N N 443853 highway bus_stop {(\\name\\,\\Neumarkt\\)} 100 \N \N \N \N \N \N \N \N \N SRID=900... Error occurred, cleaning up PostgreSQL server version is 9.0 and 8.4.7. I only want to import it to use with nominatim but I can't... it is very extrange. I was looking for the operator != isn't in the database Have you any idea ? Regards. ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Problem with Nominatim
Hi, I encountered a problem with Nominatim in correctly naming the area where a place is in. For example the Dutch town Groningen. The municipality Groningen is in the province Groningen and the country the Netherlands. Nominatim says that is belongs to the districts Aurich and Grafschaft Bentheim (two different districts) and Germany (link). Yes, it seems the Netherlands country polygon was broken in the dump I used to build nominatim. It has since been fixed but there are some knock on effects around the edges of the country. I'll force an update of everything in the Netherlands and that should fix it shortly. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [Potlatch-dev] What do the new cycleway values mean?
Well the Florist icon stands out like a sore thumb to start with - not only is it a different colour to almost all the other shopping icons but it is also a completely different shape. The florist icons certainly needs to be removed or attribution needs to be added since it is sourced from here: http://code.google.com/p/google-maps-icons/wiki/License -- Brian ___ Potlatch-dev mailing list Potlatch-dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/potlatch-dev
Re: [OSM-dev] [OSM-talk] Nominatim updates
Hi, http://nominatim.openstreetmap.org/search?q=fuelnearlat=43.5nearlon=7format=xml Is there any plan to bring this kind of (very useful !) search back ? I've fixed this - a column had changed name - thanks for the bug report. BTW - you may find that using the new bounded=1 option provides more control: http://nominatim.openstreetmap.org/search?q=[fuel]viewbox=-1.54%2C53.41%2C-1.43%2C53.36bounded=1 bounded=1 limits the search to only items in the viewbox the [] around fuel force it to be interpreted as a tag search (so it won't find places called 'fuel') Hope this helps. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] [OSM-talk] Nominatim updates
On 28 January 2011 14:47, Nic Roets nro...@gmail.com wrote: For me, reverse lookups are completely wrong right now. (I haven't tried forward lookups). And the status indication on nominatim.openstreetmap.org is blank. So I figure he's doing a DB rebuild or something. http://osm.org/?lat=-25.797306lon=28.289618zoom=18 Thanks for reporting this - this should also now be fixed. It was finding the nearest road with houses on rather than just the nearest road - a bit of an edge case this one because the road you wanted doesn't have houses but all the ones nearby do. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] type keyvalue[] does not exist error at Nominatim import
Hi, osm2pgsql -lsc -O gazetteer -C 6144 -U gis -d asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/gazetteer-functions.sql | psql -U postgres gazettee You ran the two commands on different databases. -- Brian On 25 January 2011 18:38, fatzopilot fatzopi...@gmx.net wrote: This has been discussed before, e.g.: http://www.mail-archive.com/geocoding@openstreetmap.org/msg00112.html My setup is Postgres 9.0 with postgis 1.5.2 on Ubuntu Lucid (10.04) 64bit, osm2pgsql SVN version 0.70.5 (the very recent) The following steps have been taken (according to http://wiki.openstreetmap.org/wiki/Nominatim/Installation) to set up a Nominatim/Gazetteer stack: createdb -U postgres asiagazetteer createlang -U postgres plpgsql asiagazetteer psql -U postgres -d asiagazetteer -f /usr/share/postgresql/9.0/contrib/pg_trgm.sql psql -U postgres -d asiagazetteer -f /usr/share/postgresql/9.0/contrib/_int.sql psql -U postgres -d asiagazetteer -f /usr/share/postgresql/9.0/contrib/postgis-1.5/postgis.sql psql -U postgres -d asiagazetteer -f /usr/share/postgresql/9.0/contrib/postgis-1.5/spatial_ref_sys.sql psql -U postgres -d asiagazetteer -f /usr/share/postgresql/9.0/contrib/hstore.sql psql -U postgres -d asiagazetteer -c 'create function execute(text) returns void as $BODY$BEGIN execute $1; END;$BODY$ language plpgsql;' psql -U postgres -d asiagazetteer -c 'select execute('''ALTER TABLE ''' || tablename || ''' OWNER TO gis;''') from pg_tables where schemaname = '''public''';' nohup time osm2pgsql -lsc -O gazetteer -C 6144 -U gis -d asiagazetteer /home/osm/planet/asia.osm.pbf asiagazetteerplanet.txt The last command resulted in the following output: Using projection SRS 4326 (Latlong) NOTICE: table place does not exist, skipping NOTICE: type keyvalue does not exist, skipping NOTICE: type wordscore does not exist, skipping NOTICE: type stringlanguagetype does not exist, skipping NOTICE: type keyvaluetype does not exist, skipping NOTICE: function get_connected_ways(pg_catalog.int4[]) does not exist, skipping Mid: pgsql, scale=1000, cache=6144MB, maxblocks=786433*8192 Setting up table: planet_osm_nodes NOTICE: table planet_osm_nodes does not exist, skipping NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index planet_osm_nodes_pkey for table planet_osm_nodes Setting up table: planet_osm_ways NOTICE: table planet_osm_ways does not exist, skipping NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index planet_osm_ways_pkey for table planet_osm_ways Setting up table: planet_osm_rels NOTICE: table planet_osm_rels does not exist, skipping NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index planet_osm_rels_pkey for table planet_osm_rels Reading in file: /home/osm/planet/asia.osm.pbf Processing: Node(73161k) Way(3319k) Relation(29190) parse time: 817s Node stats: total(73161946), max(1015044155) Way stats: total(3319583), max(87298623) Relation stats: total(29193), max(1298962) node cache: stored: 73161946(100.00%), storage efficiency: 15.17%, hit rate: 100.00% Stopping table: planet_osm_nodes Stopping table: planet_osm_ways Stopping table: planet_osm_rels Building index on table: planet_osm_ways Building index on table: planet_osm_rels Stopped table: planet_osm_nodes Stopped table: planet_osm_rels Stopped table: planet_osm_ways osm2pgsql SVN version 0.70.5 434.63user 4.88system 27:40.23elapsed 26%CPU (0avgtext+0avgdata 15474496maxresident)k 1429032inputs+1432outputs (74major+1003150minor)pagefaults 0swaps Afterwards, following commands have been applied (aacording to the installation instructions): cat /home/osm/bin/osm2pgsql/gazetteer/import_country_osm_grid.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_worldboundaries.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_country_name.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_gb_postcode.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_gb_postcodearea.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_us_state.sql | psql -U gis asiagazetteer cat /home/osm/bin/osm2pgsql/gazetteer/import_us_statecounty.sql | psql -U gis asiagazetteer //this is to make the sql point to the correct gazetteer.so path sed -i 's_home/twain/osm2pgsql/gazetteer/gazetteer.so_/usr/local/lib/osm2pgsql/gazetteer.so_' /home/osm/bin/osm2pgsql/gazetteer/gazetteer-functions.sql cat /home/osm/bin/osm2pgsql/gazetteer/gazetteer-functions.sql | psql -U postgres gazetteer The last command returns: CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION ERROR: type keyvalue[] does not exist CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE FUNCTION CREATE
Re: [OSM-dev] type keyvalue[] does not exist error at Nominatim import
Alas, there is still this place_boundingbox error. Can it just be ignored? You need to run gazetteer-functions.sql, then gazetteer-tables.sql, then gazetteer-functions.sql again. The second time there will be no errors. Unfortunately they have mutual dependencies. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Nominatim updates
Hi, On Monday morning I'm intending to switching the existing nominatim service over to an updated code base designed to allow better scaling and additional data sources (tiger, external postcode sets, etc) as well as being the first stage of various other improvements. With any luck the only externally visible difference should be in improvement in US search quality, however it is possible that there will be some other minor differences. If you spot any problems please email me or contact me on IRC (twain47). In particular please let me know if you spot something that used to work and now doesn't. The new service is currently a couple of weeks behind on data but catching up rapidly - I'm expecting it to be fully up to date by the end of the week and once it gets there is should be able to handle updates more quickly and reliably. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Nominatim Special Phrases and improving search
Hi, I'm changing how Nominatim handles the special phrases which are used to search for particular key=value pairs. Up until now these phrases have been imported from the translatewiki but this is now turning out to be to limited - in particular regarding the ability to capture the differences between phrases that should search for things 'near' somewhere, or 'in' somewhere or 'named' something. There are also problems with capturing the different plural version of words and the different language forms. As a result of all this I've now moved the these phrases into the wiki as a set of tables - one table/page per language. The base page (which links to all the currently languages) and contains a full explanation can be found here: http://wiki.openstreetmap.org/wiki/Nominatim/Special_Phrases In a lot of cases these pages have been machine generated with very little understanding of the particular languages and will contain numerous errors and omissions so at this point I'm now looking for volunteers to go through these pages and fix them up, or indeed to add completely new languages. I'm afraid its a fairly boring and repetitive task - I should know I've been correcting the English language version myself - but it is something which can result in a a significant improvement of the search system. Please feel free to copy or translate this message onto any of the other mailing lists. Many thanks for any assistance! -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] New public nominatim service (mapquest)
There has already been an announce of talk regarding this but I thought I'd bring this to the attention of the dev list especially given the current network problems. Mapquest have now created an instance of nominatim running on their network infrastructure (which I've been involved in setting up). This instance is rather more powerful than the instance running on nominatim.openstreetmap.org and includes redundant hardware - so although there is no actual service level commitment the hope is to provide a 24/7 service level. There are also no usage limits and the machines should be able to handle a much higher load than the existing server so I encourage you to switch particularly if this is a mission critical service for you. If you want to try it just replace: http://nominatim.openstreetmap.org/ with http://open.mapquestapi.com/nominatim/v1/ the 2 instances should be completely equivalent, same parameters, same results. Any problems let me know... -- Brian (nominatim developer) ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Postgres Error with gazetteer database for nominatim
Nothing you missed - something I missed. I have once again (I really need to stop doing this!) committed half a set of changes. To get you back working just do: alter table word add column operator text; null values default to the current behaviour. -- Brian On 11 October 2010 15:22, Lück, Michael michael.lu...@bonk-consulting.de wrote: Hi all, we set up our own nominatim server as described on http://wiki.openstreetmap.org/wiki/Nominatim/Installation . We took latest revision of osm2pgsql from the svn http://svn.openstreetmap.org/applications/utils/export/osm2pgsql. Everything seems to work until we try to search for an city or something. Then we get the following error: „nativecode=FEHLER: Spalte »operator« existiert nicht at character 71“ In Englisch: „nativecode=ERROR: Column operator doesn’t exist at character 71“ The SQL Statement fired against the database is select word_id,word_token, word, class, type, location, country_code, operator from word where word_token in (' ilmenau','ilmenau',' germany','germany'); I had a look at the table world and it really does not contain an column called “operator”. Did we miss something? Regards, Michael ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Help with implementation of multiplicatively weighted Voronoi diagram for nominatim
Hi, I've been trying to get a working implementation of a multiplicatively weighted Voronoi diagram written now for nearly a week and I'm really struggling. The intention is to use it to improve the the indexing quality and speed of nominatim with regards to mixing city, town and village points in some layers - I'm sure many of you have noticed the current problem were towns and villages end up inside city boundaries (producing weird addresses). I have a working implementation for a non-weighted algorithm using Fortune's algorithm [1] - if anyone has the time and maths skills to adapt that it would be wonderful (can Fortune's algorithm even do multiplicatively weighted Voronoi diagrams?) beyond that I've been looking at adapting the demo from cgal [2] but I'm struggling due to my poor C++ skills (and the fact that the c++ code makes use of insane numbers of templates). For someone who is really good with c++ or already familiar with cgal it would probably be fairly easy. Alternatively if anyone is aware of any other implementation or is able to implement anything based on a different library that would also be good. I think it really has to be c or c++ - anything else would be tricky to integrate. Potentially an implementation in R [3] using the postgresql module is another possibility. If anyone can help let me know - otherwise I will struggle onwards and hope to get somewhere! -- Brian [1] http://en.wikipedia.org/wiki/Fortune's_algorithm [2] http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Apollonius_graph_2/Chapter_main.html [3] http://www.r-project.org/ ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Help with implementation of multiplicatively weighted Voronoi diagram for nominatim
beyond that I've been looking at adapting the demo from cgal [2] but I'm struggling due to my poor C++ skills I hate to poison dev with anything pertaining to licenses, but note that the CGAL code you refer to is licensed under QPL which is not compatible with GPL, so if you did incorporate their code/algorithm you'd have to release the affected bits, or the whole of Nominatim, under that QPL. The license doesn't effect data generated by code - only the code itself. As such my plan was to build it as a separate module (licensed QPL) which should provide enough separation to avoid having to relicense the rest of the code base. I hope my reading is correct. However the license does make me prefer an alternative solution I just haven't found one. The alternative is to add the text given here: http://en.wikipedia.org/wiki/Q_Public_License -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] patch for nomatim
On Sun, Mar 21, 2010 at 2:51 PM, jamesmikedup...@googlemail.com jamesmikedup...@googlemail.com wrote: I have problems with the current nomatim code to connect to the database. http://trac.openstreetmap.org/ticket/2813 You made an IRC request, a mailing list post, a trac bug report [1], a comment on your blog [2] and one on your diary [3]. All of these state that the database code didn't work and that you had to 'hack' it when in fact it works fine. You added an extra / to your DSN, that is the whole problem. Better yet the comment in your own test script[1] gives the reason for your problem if you look closely. Please stop bad mouthing other peoples code - or at least make sure it is true first. -- Brian [1] http://trac.openstreetmap.org/attachment/ticket/2813/test.php [2] http://osmopenlayers.blogspot.com/2010/03/nomatim-kosovo-and-albania-running.html [3] http://www.openstreetmap.org/user/h4ck3rm1k3/diary/10067 [4] http://trac.openstreetmap.org/attachment/ticket/2813/test.php ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Nominatim json -- coment
/* Data Copyright OpenStreetMap Contributors, Some Rights Reserved. CC-BY-SA 2.0. */ This content needs to be in the file. Feel free to suggest alternative ways to format or present it but I don't think straight removal is possible. This breaks one of our tools that uses the php json_decode function, as comments are not part of the json spec. I'd like to ask politely to remove this comment. I'm uncertain why the json version would be parsed by anything but javascript - why not use the xml version? -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Nominatim json -- coment
This content needs to be in the file. Feel free to suggest alternative ways to format or present it but I don't think straight removal is possible. I doubt that it needs to be in the file; the OSM API hands out OSM data without such comments and has worked quite well for a few years now! I've added it because people are using the api to geocode whole databases and I wanted to make sure they were fully aware of the license the data was under. Above seems like the 0.6 API is isn't compliant with CC-By-SA? If you feel it must be there, is there a JSON-compliant way to encapsulate it in a variable inside the structure returned? I have moved it into a property (which should be compliant but is slightly larger download). -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Nominatim json -- coment
Because we already developed a tool that bases on the json response and now it's broken. Why do you offer the json response type if you don't deliver json? this is neither javascript (it's lacking an assignment operator) nor json, so even the Mime-Type it's delivered with is not correct. Just at the moment I'm at a loss to know why I spend my free (unpaid) time developing a tool when a small mistake gets me a load of abuse. I don't use the JSON output - I've provided it because people requested it. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Nominatim json -- coment
Don't be too upset. I believe Peter would have simply fixed it himself but felt it more appropriate to raise the issue on the list first. I have now commented out the offending bit in the code in SVN (but of course it needs to be deployed on the site to come into force). Peter, since you seem to be familiar with the JSON output, can you suggest a method to put the attribution in the file without offending the JSON spec? I have fixed and deployed already. ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] phonetic gazateer, would be cool feature for nominatim
On Fri, Jan 15, 2010 at 2:09 AM, Brian Quinion openstreet...@brian.quinion.co.uk wrote: Lucene uses Double Metaphone[1] (amongst others) and it works really nice across multiple languages. Certainly the best I've seen so far and the one OSMdoc is using in the next version. A bit of experimenting later I've made this live: http://nominatim.openstreetmap.org/?q=sheffeild however it is far from fool proof both because double metaphone isn't great with typos - only with phonetic spelling - and because the size of the database makes for a very large number of potential hits. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] phonetic gazateer, would be cool feature for nominatim
this is usually handled by using a specific language spelling = phonemes translator. so that would need a setting for 'language of the destination' which can be found automatically if a country name can be found in the text Language of the destination is indeterminate because they could be searching for the place name as expressed in another language i.e. Llundain (welsh version of London) so even if you can determine the country it doesn't work. And the similar in reverse for the browser language - they could be searching for London even though they are a native German speaker. Also it isn't the phonetic spelling that is really the problem - the problem is that a lot of the time mistakes are typos not miss-spellings i.e. Lonon which is phonetically completely different to London despite being obvious that someone just missed the 'd' when they were typing. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] osm2pgsql deltas
8GB of RAM. Were you using slim mode? I can import in under a day when not using slim mode, however you have to use slim mode if you want to be able to apply deltas and this adds a huge overhead. Did you increase the cached size to allow osm2pgsql to use more memory (--cache)? The default is only 800 MB and with your machine you should be easily able to allocate 4000 MB or so - I found this had a massive effect and is easily missed. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] phonetic gazateer, would be cool feature for nominatim
On Thu, Jan 14, 2010 at 4:57 PM, Mikel Maron mikel_ma...@yahoo.com wrote: In the Haiti response, someone asked about the Location of Lilavois. That gets no results on most gazateers, except this one http://dma.jrc.it/services/gazetteer/?q=lilavoiscc=HA The place is actually called Lillavois (two L's). This gazateer searches phonetically. I'd love to do something like this but have so far been unable to find a suggest / phonetic algorithm that scales to 5.4 million words/phrases across more than a hundered languages. The geocoder you mention doesn't seem to do street level - which simplifies the problem by at least a couple of orders of magnitude. If someone can provide an algorithm I'm even happy to code it but so far I've had no luck in finding any way to do this. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] How to support name:* in MapOSMatic?
On Fri, Jan 8, 2010 at 12:57 PM, David MENTRE dmen...@linux-france.org wrote: The ideal configuration would be to have all names (name:fr, name:ar, ...) into a single name column and use it. Or have a kind of couple (name, country_code) that would store all different names for a street. But I don't know how to do this. That is how the gazetteer output from osm2pgsql handles it already. It creates a type of keyvalue as: CREATE TYPE keyvalue AS ( key TEXT, value TEXT ); and then has a column of type keyvalue[] (keyvalue array) All name values are then merged into this field during the osm2pgsql import. It ends up with a field value like: {(name:en,Peschanka),(name,Песчанка),(alt_name,Pesnanka)} Then a function is used to retrieve the name in the required value by specifying a preference order: CREATE OR REPLACE FUNCTION get_name_by_language(name keyvalue[], languagepref TEXT[]) RETURNS TEXT AS $$ DECLARE search TEXT[]; found BOOLEAN; BEGIN IF (array_upper(name, 1) is null) THEN return null; END IF; search := languagepref; FOR j IN 1..array_upper(search, 1) LOOP FOR k IN 1..array_upper(name, 1) LOOP IF (name[k].key = search[j] AND trim(name[k].value) != '') THEN return trim(name[k].value); END IF; END LOOP; END LOOP; RETURN null; END; $$ LANGUAGE plpgsql IMMUTABLE; select get_name_by_language(namearray, ARRAY['name:en','name:de','name']) from table; -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Major improvements to MapOSMatic
Vector (SVG) symbols in mapnik are a work in progress, and not currently implemented in the main codebase. Are higher resolution icons an option? The majority of the icons used are available as SVG, see here: http://www.sjjb.co.uk/mapicons/SJJBMapIconsv0_03.tar.gz (if anyone can ever get the svg icons working in mapnik!) In the mean time larger png versions are available here: http://www.sjjb.co.uk/mapicons/SJJBMapIconsv0.03/recoloured/ If you need even bigger ones I'll happily render them, or there are scripts included with the SVG files to generate them. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Configurable number of results in Nominatim ?
On Wed, Dec 23, 2009 at 6:32 PM, Thomas Petazzoni thomas.petazz...@enix.org wrote: However, the current version of Nominatim limits the number of results to 10. While this makes perfect sense for the Web application, it isn't really pratical for the XML export. For a lot of purposes the limit is fine even with xml or json - it gives back the (up to) 10 mostly likely matches. Perhaps if you could give a use case to explain why you would like this limit removed I could suggest other options. Is it possible to add support for a limit=XX argument, where XX is the number of results required, with 0 being a special value removing the limit ? The limit of 10 items is not a display choice but a performance choice. It is there to prevent a single query for using too much resources and blocking usage by other people and to the ensure that the response is returned in a reasonable time. Have you investigated the possibility of using the more_url ? This already allows you to obtain the next batch of 10. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] JSON-Decoding-Bug in Nominatim
Ich habe lese mit PHP Daten von der nominatim API via JSON. Das Query http://nominatim.openstreetmap.org/?q=Stadeformat=json gibt mir ua. folgende Property aus: display_name:Stade, Avenue de Barbe d\'Or, Mont-de-Marsan, Landes, Station Total, Aquitaine, Frankreich See if that works for you now. BTW - on a related note I've just added various extra JSON(P) options to Nominatim. Documentation here http://wiki.openstreetmap.org/wiki/Nominatim -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] [PATCH] for mod_tile, adds config file support and support for multiple tile sets
Hi, The patch is too large for the mailing list, so you can instead download it here: http://osm.sjjb.co.uk/mod_tile_ini.diff This patch adds various extra options to mod_tile: - Support for multiple named mapnik/tile configurations - Extra apache directives, e.g.: AddTileConfig /uri/ ConfigName AddTileConfig /osm/ Postcode LoadTileConfig /etc/renderd.conf - Config file support (by default /etc/renderd.conf) of the format: [Default] URI=/osm_tiles2/ XML=/home/jburgess/osm/svn.openstreetmap.org/applications/rendering/mapnik/osm.xml [Postcode] URI=/osm/ XML=/home/jburgess/osm/svn.openstreetmap.org/applications/rendering/mapnik/postcode.xml - Changes so mod_tile only effects a specific apache virtual host (the one where the Add/LoadTileConfig directive is used) - Removed the requirement for the meta tiles directory to be in the web root, moved by default to /var/lib/mod_tile/ As a side effect of these changes the protocol version is moved up to 2 and an extra level of hierarchy is introduced into the HASH_PATH for the name of the mapnik xml file. To preserve the existing meta files they need to be manually moved to the new folder structure by creating a 'Default' folder and moving the files into it. I've tested this but could easily have missed something. If people are happy with these changes I will happily commit them to svn but it's quite a significant change! -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Request for UK address lists for postcode extraction
Hi, I'm currently doing some work trying to generate postcode location data for the UK using address lists and address lookup using OSM data to supplement NPE. So far it seems to work quite well with the address lists that I have available to me (and coping quite well with ambiguous road names) but I'm limited in my data sources and most of the address data is fairly consistent in both format and quality. So, before I open the interface to the public, I'd like to test the code with some lists provided by other people. Does anyone have, or know of, any address lists that I would be able to use for this purpose? Obviously it needs to be license compatible with OSM (so please no lists generated from royal mail postcode data!) and ideally I'm after data sets containing at least: street address (house name / number optional) town / city postcode formatted as CSV or TSV. I'm specifically not after data containing the names of individuals. Has anyone got any suggestions, or is willing to offer any data? Even personal address books would be useful for testing... Cheers, -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] 'holes' in buildings
I'm running my own mapnik server and have noticed a strange problem with the rendering. I don't seem to be getting the holes in buildings You need to use osm2pgsql slim mode (-s), apparently relation parsing is broken in the in-memory mode, since it can't decide which ways are valuable enough to keep before parsing the relation. (Or something similar) Running the current SVN version of osm2pgsql in slim mode does fixed the problem. Many thanks! -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [josm-dev] JOSM extension for orthogonalizing ways
This is my big problem with the current implementation - it creates a rectangle in the current projection with the result that (as you go further north / south) the rectangle become more and more misaligned to physical reality. If you use the Mercator projection that problem should be less pronounced. Yup - I switched a while back and it improves things a lot - although it is still out by a bit. It is not easy to do rectangular alignment on the surface of a sphere which is basically what you're asking for. The geometry code is complicated already even if you're just dealing with projected data! Indeed! This is the reason I've not tried to fix it myself - it makes my head hurt :-) We could of course simply disable the option for the EPSG:4326 projection. Which used to be the default because of the inability of the WMS plugin to work properly with anything else. Maybe the new WMS code doesn't break as bad when used with Mercator, in that case we could switch the default to Mercator. Disabling or warning the user in EPSG:4326 would seem a major improvement. It just worries me that at the moment the tool makes data less accurate in the default projection - catching out new people and those least likely to realise there is an issue. I'm assuming from this that the new code from Harald doesn't resolve this? As I said not actually read it to check... Cheers, -- Brian ___ josm-dev mailing list josm-dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/josm-dev
Re: [OSM-dev] westbank mapnik rendering, icons not showing. help?
Mikel, Just a guess but see if you have a Thumbs.db in svn.openstreetmap.org/applications/rendering/mapnik/symbols If you do try deleting it and doing another svn update and see if you get any extra updates I've just hit this - svn bailed due to a conflict and I didn't notice. -- Brian On Thu, Oct 16, 2008 at 3:37 PM, Mikel Maron [EMAIL PROTECTED] wrote: I reinstalled mapnik and mod_tile yesterday on http://westbank.maps.jsintl.org/ Looks great! but this particular install has a problem .. POI symbols are not showing up at all. Has anyone encountered a similar problem? Solution? Or, any ideas on how to go about debugging? Thanks Mikel ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [josm-dev] Preset Icon Backgrounds - SCNR :)
Hi, I'm glad you found the icons helpful. Just to say that the last version is now in svn (http://svn.openstreetmap.org/applications/share/map-icons/svg-twotone/) and the version on the sjjb.co.uk website is out of date. The version in svn contains quite a lot more icons and various improved scripts for generating png icons - generatemapniksymbols.sh and generatecategoryicons.sh most likely already do what you need. -- Brian On Thu, Oct 2, 2008 at 12:11 AM, Christoph Eckert [EMAIL PROTECTED] wrote: Hi, more or less by accident, I tried to give some of the preset icons a transparent background. To achieve this, I copied over some icons from mappaint first, as mappaint icons often need a background color due to the [default] black background in JOSM. I converted many icons to BW, as black is beautyful - just kidding. I got some momochrome icons I found much more readable at 16x16 than the previous ones. I might be wrong, though. I used some PD icons as found on commons.wikipedia.org, and some from http://www.sjjb.co.uk/mapicons/ The latter source contains a converter script http://www.sjjb.co.uk/mapicons/SJJBMapIconsv0.01/generateicons.sh which simply uses sed to derive various color schemes from the svg files. I wonder if we could use a similar script to derive the preset and mappaint icons from the same source automatically instead of doing it manually. Needs some consideration about the file naming and folder structure. I'm for a completely flat structure in various folders for various styles. The script then could derive subfolders for the different styles (White on white, Blue on blue, Black on black and so on). Just my two cents, ce ___ josm-dev mailing list josm-dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/josm-dev ___ josm-dev mailing list josm-dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/josm-dev
Re: [OSM-dev] [Patch] Experimental extra symbols for mapnik layer
sorry - long post, verging on brain storming... I seem to have misunderstood something, and thought someone suggested having multiple layers of POIs. Slowing down to type and perhaps that suggestion made, sorry if it's an outrageous one. But your icon set looks like it could be easily split into a few layers, brown layer(food/pubs) etc. I also really like the transparency on the beer icons and generally. While having lots of layers in great in theory I think that there are several practical problems to do with both server side storage, increases in data transfer, client side rendering speed and overloading the interface with too many options. You can probably see the client side rendering speed problem fairly easily on the dutch cycle site that Rob provided a link to: http://www.openfietskaart.nl/?zoom=14lat=52.36713lon=4.89568layers=BFTTT turn on 3 or more layers and then try moving the map - the extra time to download 3 times as many tiles adds a noticeable extra latency (on my system anyway). The storage problem and bandwidth issues would be similar although more easily solved / ignored. There is also the problem of generating layers in such a way that they don't overlap - tricky because mapnik isn't really setup to taken account of things that are not rendered - so I think that probably everything (except the icons) would have to be rendered using an alternative transparent style sheet so that the icons ended up in the right place. Urrrg. Of course we could just ignore the overlapping problems :-( The alternative is to do the poi client side like: http://www.lenz-online.de/cgi-bin/osm/osmpoinit.pl/ (which unfortunately seems to be broken at the moment, for me anyway) Again there is the overlap problem, made even more difficult by being client side. Client side latency is still a problem for loading the extra icon location data, as is over complicating the interface. There is also a new problem which is that icons for features that are already rendered by on the mapnik layer get in the way / are redundant. Layers just seem to be tricky... so we are left with the current mapnik / osm.xml seems to be a compromise between all the various options that showcases what can be done with the data to produce a nice map. Unfortunately it suffers from the standard problems with all compromises - some people want more detail/icons (like myself) some people want less (I remember a post requesting a version without pub symbols). And at the moment rolling your own mapnik server / osm.xml is not exactly trivial especially if you want extra icons not in the standard definition. (as an aside, is direct usage of the osm title servers (i.e. c.tile.openstreetmap.org) by other projects/organisations encouraged or discouraged? I'm never quite sure and it makes a difference as to the purpose of the mapnik layer...) Personally I think that the solution might be 2 versions of the mapnik layer: mapnik-min and mapnik-max (preferably both generated from the same template). Of course this assumes that the server has the resources to cope with caching 2 sets of tiles (does it?) and that anyone except me things this is a reasonable idea! Such a move might also be basis for trying to find some way to split up the existing osm.xml which as Steve Chilton mentioned is now well over 5000 lines and finding a way to make it easier to customise osm.xml for roll your own maps without having to branch off completely on your own. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] [Patch] Experimental extra symbols for mapnik layer
On a related note - does anyone know who admins the live server, or how changes to the mapnik style sheet are made live? Is there someone in charge or is it just a case of committing the changes to svn and it all happens automatically? -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] [Patch] Experimental extra symbols for mapnik layer
Steve Chilton is cartographer-in-chief/benevolent dictator for the Mapnik stylesheet - that's how it keeps looking so good. See http://trac.openstreetmap.org/log/applications/rendering/mapnik/osm.xml Best discuss any proposed changes with him before committing. Ah - sounds like I guessed right to start with, unfortunately I emailed him a week ago and got no response... my suspicion is that my email is probably sat it a spam folder somewhere. So Steve, if you are out there please get in touch! :-) -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] [Patch] Experimental extra symbols for mapnik layer
On Wed, Sep 17, 2008 at 4:19 PM, Richard Fairhurst [EMAIL PROTECTED] wrote: On a cartographical note: the icons look good and consistent, and work well per se. Cheers! However, IMeversoHO the result of adding this number of icons - certainly not a reflection on your icon design - is that the map becomes a little too cluttered. It loses the clean look of our default Mapnik cartography and that, I think, is a shame. The automatic placement of point labels over ways tends to 'break up' the ways, making them harder to comprehend at a glance: Surrey Street in your osm-sample2.png is a good example. Actually I agree! Adding icons need to be done careful moderation and adding any significant number runs the risk of making the default map unsuitable for some purposes (for instance a hotel chain would not want other hotels appearing on their maps!). I'm also not particularly happy with the placement of the icons - I'm currently working on a patch to mapnik to allow icons to be 'nudged' into clear space to improve the cartographical result - I'm also trying translucency and various other effects. Suggestions welcome. Stefan de Konink alluded to the ideal solution in his posting - just another reason to have 'layers'. :). If these icons weren't visible by default, but were available as a selectable overlay, that'd be utterly fantastic. This sounds like a good alternative - but if anything it involves more work and more server side changes than just adding the symbols (which is quite a bit of work in itself!). I'm happy to work on something like that - but only if I know it is going to be used at the end so this is in the way of a feeling out to see what people think. I can also think of other alternatives too. For instance we could provide 2 mapnik layers (one naked, one with icons) or try to find some way to overlay individual symbols browser side without too much of a performance hit. -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] [Patch] Experimental extra symbols for mapnik layer
Hi, I've created a patch to add extra amenity and tourist icons to osm-template.xml from the svg-twotone set of shared symbols that I recently added to svn. The hope is to produce a consistent style so this patch not only adds new symbols but also replaces some of the existing ones. I've not yet added all the symbols from the icon set to the osm.xml file but I'd like to get some feedback on what people think before I spend too much more time on this. I've produced a couple of screen shots to give a quick impression of what the results looks like. You can see them here: http://www.sjjb.co.uk/mapicons/osm-sample1.png http://www.sjjb.co.uk/mapicons/osm-sample2.png The two samples only include a few of the icons added but should give a general idea. Alternatively if anyone has somewhere I could put a demo copy of the config for people to look at that would be very helpful - my local development server just isn't up the job! The patch to osm-template.xml is available here: http://www.sjjb.co.uk/mapicons/osm-template-map-symbols.patch In addition to applying the patch you will also need to run the generatemapniksymbols.sh script in applications/share/map-icons/svg-twotone to generate the required png files. Please do let me know what you think! Cheers, -- Brian ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev