On Mar 23, 2011, at 9:40 PM, Jin Kang wrote: > Hi Dane, > > Following up previous conversation, we were able to compile successfully but > now we have another question. > > 1. For mapnik2 update, do we need to install mod_tile?
If you were previously running mod_tile/renderd, then yes you need to recompile. To recompile and link against mapnik2 you need to 1) stop any running versions of renderd 2) uninstall your previous mapnik installation's headers 3) install mapnik2 (or ensure it is installed correctly) 4) change the mod_tile Makefile to link to -lmapnik2 Details here: http://trac.mapnik.org/wiki/Mapnik2#mod_tilerenderd > > 2. If so, when we execute the command 'make' as below, > > -------------------------------------------- > # build mod_tile/renderd with above patch > > svn co http://svn.openstreetmap.org/applications/utils/mod_tile > cd > mod_tile > make > # will take a long time Yep, should take a long time and requires > 1.5 GB RAM. Mapnik2 takes 2-3 longer to compile against (slower compiles == faster runtime). > > -------------------------------------------- > > We are getting following error: > > Makefile:5: *** "cannot find apxs or apxs2". Stop. That is unrelated to anything with Mapnik. Sounds like you a missing the apache development headers/programs. Or, if you are running on the same machine as mod_tile previous compiled on, maybe you modified something in the Makefile that caused a syntax error. Makefiles require true tabs I think - so make sure you are using tabs for indents. > > We are not sure what causes this error so your comments will be appreciated > as usual. Ya, I don't know, other than to make sure apxs is installed, which is an apache thing. Google it. > > Thanks much! > > -Jin > > On Mar 16, 2011, at 5:33 AM, Dane Springmeyer wrote: > >> Jin, >> >> Looks like maybe your boost version is old. What version did you install? >> Also when you ran: >> >> python scons/scons.py configure >> >> What version of boost does it report finding? >> >> Dane >> >> >> On Mar 15, 2011, at 12:54 AM, Jin Kang wrote: >> >>> Hi Dane, >>> >>> We tried to compile but getting following error during mapnik2 compilation. >>> >>> ------------------------------------------------------------------- >>> src/svg_parser.cpp:122: instantiated from here >>> /usr/local/include/boost/spirit/home/qi/detail/pass_container.hpp:236: >>> error: no type named ‘type’ in ‘struct >>> boost::spirit::traits::container_value<std::pair<std::basic_string<char, >>> std::char_traits<char>, std::allocator<char> >, std::basic_string<char, >>> std::char_traits<char>, std::allocator<char> > >, void>’ >>> /usr/local/include/boost/spirit/home/qi/detail/pass_container.hpp:248: >>> error: no type named ‘type’ in ‘struct >>> boost::spirit::traits::container_value<std::pair<std::basic_string<char, >>> std::char_traits<char>, std::allocator<char> >, std::basic_string<char, >>> std::char_traits<char>, std::allocator<char> > >, void>’ >>> scons: *** [src/svg_parser.os] Error 1 >>> -------------------------------------------------------------------- >>> >>> These are the steps we took to install Mapnik: >>> >>> 1.Building ICU → installation was okay >>> 2.Building Boost → installation was okay >>> 3.Getting Mapnik2 source → okay >>> svn co http://svn.mapnik.org/trunk mapnik2 >>> 4. Mapnik installation >>> rm -rf /usr/local/include/mapnik >>> # re-install mapnik2 headers >>> >>> cd >>> <mapnik2 sources> >>> sudo python/scons/scons.py install # → This is the line we are getting >>> above error. >>> >>> >>> Any thoughts on this? Thanks in advance! >>> >>> Regards, >>> >>> -Jin >>> >>> >>> On Mar 14, 2011, at 7:01 AM, Dane Springmeyer wrote: >>> >>>> Jin, >>>> >>>> I've just imported osm for japan to help get set up for producing products >>>> for the quake response. >>>> >>>> I've noticing the really large relations that Lennard called out as the >>>> problem. I think he was spot on about this being the problem. I have also >>>> connected with Artem who appears to have greatly improved Mapnik's ability >>>> to handle polygons of this size and complexity. >>>> >>>> So, I have documented Artem's fix at this ticket: >>>> http://trac.mapnik.org/ticket/728 >>>> >>>> What this means is that Mapnik 2.0 (current trunk) may be *much* more >>>> suited for rendering Japan OSM data. To upgrade to Mapnik trunk see this >>>> page: >>>> >>>> http://trac.mapnik.org/wiki/Mapnik2 >>>> >>>> To get your current osm.xml working with Mapnik2 you will need to upgrade >>>> it like: >>>> >>>> $ upgrade_map_xml.py osm.xml osm2.xml >>>> >>>> And then use the osm2.xml. >>>> >>>> Get in touch if you have any questions or need help. I apologize that I >>>> did not see this connection sooner. >>>> >>>> Dane >>>> >>>> >>>> On Feb 24, 2011, at 10:42 PM, Jin Kang wrote: >>>> >>>>> Sorry everyone! The following time should be minute, not sec. >>>>> amenity-poly: 5.7 min >>>>> leisure: 1.44min >>>>> text-poly: 1.43min >>>>> >>>>> Sorry for the confusion! >>>>> >>>>> Regards, >>>>> >>>>> -Jin >>>>> >>>>> >>>>> >>>>> On Feb 25, 2011, at 3:37 PM, Jin Kang wrote: >>>>> >>>>>> Hello everyone, >>>>>> >>>>>> This is Jin again from Rainbow. My engineer is rendering Japanese map in >>>>>> zoom level 16 as I mentioned before but still we are experiencing very >>>>>> slow rendering speed. Sometime, we are noticing it take several minutes >>>>>> for just one tile. We put some code to debug and found that we are >>>>>> especially seeing very slow speed in following layers. >>>>>> >>>>>> amenity-poly: 5.7s >>>>>> leisure: 1.44s >>>>>> text-poly: 1.43s >>>>>> >>>>>> Can someone tell us what are these layers for? Is it okay to deleted >>>>>> these from OSM file and rendering? >>>>>> >>>>>> Thanks in advance! >>>>>> >>>>>> -Jin >>>>> >>>>> _______________________________________________ >>>>> Mapnik-users mailing list >>>>> [email protected] >>>>> https://lists.berlios.de/mailman/listinfo/mapnik-users >>>> >>> >> > _______________________________________________ Mapnik-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/mapnik-users
