Jan-Benedict Glaw wrote: > The second bug is more subtle. The newly introduced whitespace > seems, parsed in XML context, to result in something that Mapnik > parses wrongly. From looking at SQL traces, the offender seems to be > that Mapnik (0.5) correctly parses the source table name, creating > this diff (comparison of PostgreSQL log files):
Mapnik 0.5.x indeed has an issue with multiline SQL, fixed in r862. This might be what you're seeing with 0.5. <http://trac.mapnik.org/ticket/173> > Linking against Mapnik 0.6 resulted in about an empty image (only > background, no SQL statements triggered, no C++ish exception...) I'll > try to look into that soon. So it would probably be nice to not place > a table name to the end of a line for some time... Mapnik on tile.openstreetmap.org doesn't seem to have a problem with it. Neither does my dev server. Which Mapnik revision did you compile? On #mapnik, we've also been exploring the need to be able to define the minimum Mapnik version/revision that will be able to parse all features used in a stylesheet. I hope that, given that this multiline SQL vs Mapnik 0.5.x has turned into a FAQ, this could be implemented sooner rather than later. That would prevent further such issues in the future. <http://trac.mapnik.org/ticket/395> -- Lennard _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/listinfo/dev
