[postgis-devel] [PostGIS] #2122: [raster] Real extent feature lost after metadata as views

Bborie Park dustymugs at gmail.com
Fri Nov 30 19:51:04 PST 2012


>> As I said in a previous mail this week: it is a misconception to consider a raster
>> table a raster (always regularly tiled). A raster table is more like a folder
>> containing a bunch of raster (a raster coverage) with all the unforeseeable the
>> arrangement of these many rasters may have (unalignement, overlaps, uncompletion).
>
> Pierre, please, don't do that, again.
> Don't assume you have the monopoly for valid use cases of PostGIS Raster.
> You may not like it, you may think it's a stupid idea, an unacceptable idea,
> but, let me quote Tim Keitt from his PGRaster paper:
>

Seriously, I think both of you are on opposite extremes of the same
topic.  The reality is that none of us will ever know how PostGIS
raster will be used.  So, we provide a core set of tools upon which
specific use-cases can be built upon.

> http://lists.osgeo.org/pipermail/postgis-devel/2007-July/002652.html
>
> "Blocking storage techniques, pyramid structures and data compression
> are the three key approach to improve the performance of PostGIS
> PGRaster in data processing and visualization."
>
> Blocking is a generic technique, almost equally important
> as pyramids. We can't simply ignore it.
>

Reading that thread made me laugh, especially considering the tone in
this email thread.  If anything, thorough tests need to be done on
PostGIS raster in real hardware (not virtualized) to see what
optimizations are needed to improve performance in accessing data and
analysis.

-bborie



More information about the postgis-devel mailing list