Hello Liz,
Welcome to Mapnik.
On Apr 27, 2009, at 2:26 PM, Liz Dobbins wrote:
Hi;
I've got a question that relates to Ticket #75: Feature Request: 2-
level spatial index ("TileIndex" in UMD Mapserver) - a feature that
is planned for Mapnik in the future.
I'm working for a company that has used MapServer in the past, and
is considering migrating to Mapnik to generate map tiles. As part
of the old system, there are several data sets that were configured
to use the TileIndex capability of MapServer. tile4ms was used to
break the original data set into localized shapefiles, each with its
own spatial index file (.qix, created by shptree). The filenames of
these data sets are summarized by a tile index shapefile that
includes the file pathnames as an attribute named "LOCATION". There
aren't many localized files; the world is covered by only 73 of
them. The tile index file is then referenced in the MapServer
stylesheet.
I'd like to convert this to a system that Mapnik can use currently.
I was thinking of importing all the shapefiles into a PostGIS
database,
Right, sounds good. We've not added support for a shapefile TILEINDEX
(and there are currently no active plans to do so) because most use
cases like this with Mapnik are handled nicely by PostGIS.
For reference however, Mapnik does support QuadTree indexes on
shapefiles like those produced by 'shptree'. These files have
an .index file extension and can be created by installing (done by
default) the 'shapeindex' command line tool bundled with Mapnik and
running it on a shapefile.
and doing the spatial indexing with it, but I'm a total newbie with
databases.
Ah, PostGIS is excellent, and there are a lot of resources for
community and commercial support. Also, the PostGIS documentation has
been improving rapidly in the last several months and you can find
descriptions of working with spatial and attribute indexes here:
http://postgis.refractions.net/documentation/manual-svn/ch04.html#id2767631
I think that a database view might take the place of the tile index
file, but I don't know how to set it up. Anybody have some
suggestions about how this could be done?
I would start by importing your shapefiles into a single PostGIS table
(don't worry about VIEW for now), and creating good spatial indexes
and attribute indexes on any fields you plan to apply attribute
filters upon.
Once you have benchmarked this approach you could break your tables up
a bit and set manual extents for each table in your Mapnik XML which
may be able to speed things up as well.
To help further it would be good to know your anticipated rendering
workflow. You must be planning to render map tiles dynamically? Is
your data frequently changing? If your data is fairly static, surely
the easiest way of optimizing response times is to implement tile
caching.
Thanks,
Liz
Cheers,
Dane
_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users
