[postgis-devel] [PostGIS] #497: [raster] Start testing displaying PostGIS raster with Mapserver

[PostGIS]

#497: [raster] Start testing displaying PostGIS raster with Mapserver
----------------------------+-----------------------------------------------
 Reporter:  robe            |       Owner:  robe         
     Type:  task            |      Status:  new          
 Priority:  medium          |   Milestone:  PostGIS 2.0.0
Component:  postgis raster  |     Version:  trunk        
 Keywords:                  |  
----------------------------+-----------------------------------------------

Comment(by robe):

 Jorge -- I'm actually wondering if you should have another mode.

 I guess the question is what do you need from the meta data.  Doing a
 simple distinct won't do it and would be worse because it would still
 return many records since each tile has a different upperleftx,y and
 actually doing a DISTINCT even when I get rid of all of that still takes
 like 10 seconds to read my hires table (which is a small sample of 4 tiles
 and 20,000 records) (and 2 seconds for my low res).  It gets exponentially
 worse the bigger your table.  It really should be linear according to the
 number of tiles you are pulling (not the number of records in your table).
 So I'm really much more bothered about the meta data call than the tile
 call (that aside from the use of ~ seems like as efficient as it can be
 without too much fuss and for my case my OpenLayer tile sizes match my
 block size, so it would be making separate calls anyway via mapserver)
 because I suspect that is a good chunk of where the time is going.

 I think the general use case is that people will have a folder of rasters
 that form a non-overlapping coverage like my case and bring them into the
 table.  You can't really rely on the rid # either since in most cases (at
 least for mine), I may be bringing them in piecemeal like one town at a
 time or one state at a time especially for huge things where I need to
 provision disk space but I want people to see as a contiguous layer on the
 map.

 But more often than not the table will have the same SRID -- actually it
 has to because we enforce it with the SRID constraint during add.

 Does gdal ever look in the raster_columns table.  Part of my problem is
 that because the raster2pgsql didn't add the rast column to my overviews
 -- it was simpler for me to just type in rast for column than put in an
 AddRasterColumn call.

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/497#comment:9>
PostGIS <http://trac.osgeo.org/postgis/>
The PostGIS Trac is used for bug, enhancement & task tracking, a user and developer wiki, and a view into the subversion code repository of PostGIS project.