#1713 closed defect (wontfix)
postgis extension cannot be made without raster
Reported by: | Mike Taves | Owned by: | pramsey |
---|---|---|---|
Priority: | medium | Milestone: | PostGIS 2.1.0 |
Component: | postgis | Version: | master |
Keywords: | Cc: |
Description
When building the postgres extensions (for PostgreSQL 9.1+) with ./configure --without-raster
, the postgis
extension build fails.
There are many references to "raster" and "rtpostgis" in extensions/postgis/Makefile.in
, so this is an expected consequence.
Interestingly, if --with-topology
is configured, the topology extension can be successfully built and installed in /usr/share/postgresql/9.1/extension/
. However, from psql:
postgres=# CREATE EXTENSION postgis_topology; ERROR: required extension "postgis" is not installed
There should really be a separate postgis_raster
extension built, if configured with raster support. (I'm not sure if it is too late for 2.0.0, but I'll assume it is.)
Change History (4)
comment:1 by , 13 years ago
Resolution: | → wontfix |
---|---|
Status: | new → closed |
comment:2 by , 13 years ago
Ah thanks, that clears things up a bit. I've been looking at cleaning up a few docs on the 9.1+ Extensions, so that helps put it in perspective.
comment:4 by , 7 years ago
the old link is broken and redirects to wrong spot here is the new link of what we thought years ago.
https://lists.osgeo.org/pipermail/postgis-devel/2012-January/thread.html#17136
Believe it or not, we had long painful discussions about this and in the end decided that we want raster to be part of the core postgis extension, so building the postgis extension requires you build with raster and if you don't want that you have to do it the old way without extensions.
http://www.postgis.org/pipermail/postgis-devel/2012-January/thread.html#17158
topology is separate because well it is installed in a separate schema and limitations of extensions don't allow extensions to go across schemas.
The main reason for this is it makes install easier for those who want both raster and vector functionality. The only reason against it is it adds a dependency on gdal.
In future releases of postgis, the gdal dependency may not be just limited to raster though so keep that in mind. gdal brings in a whole wealth of opportunities -- e.g the possibility of spatial foreign data wrappers that piggy back on postgis geometry. Closer integration with our loaders etc. Besides half the postgis dev team works on gdla anyway (you might become victim of the borg too Mike :) )