Re: [OSM-dev] RAM requirement for osm2pgsql without slim?
2009/12/9 Frederik Ramm frede...@remote.org: Hi, what is the current RAM requirement for running osm2pgsql on the whole planet in non-slim mode? Is 16 MB still sufficient for that, or does it require more in the mean time? I think its more like 16GB. ie more memory that any 32bit OS can actually have! Peter ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
[OSM-dev] Minute Diffs
I've been trying to get a copy of the OSM data set into a postgres database so I can extract some statical info to help with hopefully one day search etc. Broadly speaking I've been following the instructions at http://wiki.openstreetmap.org/wiki/Minutely_Mapnik except I'm skiping osm2psql and using osmisis to import the data directly. eg osmosis -q --read-change-interval --write-pgsql-change I would like to start with just part of the database due to space, and processor limitations and have preloaded the data using bzcat planet-latest.osm.bz2 | osmosis-0.31/bin/osmosis --read-xml-0.6 file=- --log-progress --bb left=-11 right=2 top=63 bottom=49 --write-pgsql-dump enableBboxBuilder=yes enableLinestringBuilder=yes nodeLocationStoreType=TempFile Is this doable what do I need to do to get the diff to import without error. (If its possible at all) If its not I'll have to find a faster computer with more space which is not the simplest job in the world (250G is currently the largest single disk in the office and it looks like the postgres database for the whole planet is 500G+ currently (including indexes etc)) If any one knows how much space I would need that might help too. Peter. ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] Minute Diffs
2009/11/11 Peter Körner osm-li...@mazdermind.de: Peter Childs schrieb: I've been trying to get a copy of the OSM data set into a postgres database so I can extract some statical info to help with hopefully one day search etc. Broadly speaking I've been following the instructions at http://wiki.openstreetmap.org/wiki/Minutely_Mapnik except I'm skiping osm2psql and using osmisis to import the data directly. I'm using osm2psql with osmosis to import the minutely replicating diffs and get I no errors. /sql/diffs/osmosis-trunk/bin/osmosis --rri workingDirectory=. --wxc - \ 2 $OSMOSISLOG | /usr/bin/osm2pgsql --append --slim --cache 1024 \ --username gis --style /sql/planet.osm/wikimedia.extended.style \ --verbose - 12 2 $PSQLLOG As osmosis is putting the changesets into osm2pgsql via pipe you should be able to plug another osmosis task in between to cut out your bbox. The only issue with that is that it thows a nasty error Task 2-bb does not support data provided by default pip sotred at level 1 in the default pipe stack. osmosis -q --read-change-interval workingDirectory=$DIR/minute -bb left=-11 right=2 top=64 bottom=49 --write-pgsql-change database='osm' user='peter' I'm now trying to import the change sets without the bb but with only part of the original data (Which probably means I'll gather new stuff outside the original square but hopefully it will not blow up in my face for a while since it has plenty of hard drive space...) Peter ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] [Talk-GB] maposmatic - update
2009/10/23 Peter Miller peter.mil...@itoworld.com Ok, so after looking a bit more carefully and reading their blog[1] it says that they are working on the performance issues associated with running a global service. Regards, Peter [1] http://news.maposmatic.org/ ___ Talk-GB mailing list talk...@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb I'm looking at this and thinking that it should not take days to import the planet into a database... I'm currently trying to import the database into postgres using osmosis (to try and do some work on geocoding) and most of the processing it being used by osmosis I'm not i/o locked, I'm processor locked. and it looks like osm2pgsql suffers from a similar feature. I'm thinking that there must be a better way... I suspect the problem lies it knowing which nodes are needed to import all the relations in an area when some of those node may be outside the area. But this should not be needed at all if your not filtering and just importing a full database. I suspect that a Sax Parser ought to be able to read the input and output Sql without needing to do much more work, (well not much more than hold the odd bit of state) Peter. ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev
Re: [OSM-dev] footway inside building: new tag-value 'indoor=yes'
2009/6/29 Claudius claudiu...@gmx.de: Am 29.06.2009 23:03, Ed Loach: My 'problem' is that this looks graphically overloaded - think even about footways in several floors! Q1. What about a new tag indoor=yes in addition to highway=footway? Q2. Renderer (like Mapnik / Osmarender) then chould suppress these footways? Alternative: new tag highway=corridor, and the routing software could then add support for indoor routing? Ed Don't forget surface=carpet/parquetry/tiles/linoleum :P Instead of reusing the plaing highway-tag I would favor a new namespace for indoor features. Sth. along indoor:highway=corridor or indoor:highway=steps The problem that springs to mind, Is shopping centres. Where all the shops are inside and the Way is the indoor bit between them. I can also see someone trying to put a road in for a multi-story car park. I think the inside=yes is a good idea. I also think the ways still need rendering if its public, As it means you can walk through the building. Maybe something like the way Landuse gets rendered. Peter. ___ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev