[postgis-users] [postgis-devel] postgis extension sans raster (only for folks who can't compile with raster support) - PSC Vote and developer/packager comments please

[Vincent Picavet (ml)]

Hello Regina,

We do have cases where we do not need raster at all (>90% of our users),
and must compile everything from scratch (~10% of our users). We would
be pleased to not having to compile GDAL in this case.
And it would be great to still use the CREATE EXTENSION mechanism to
install Postgis, just to be consistent from a user point of view, with
what can be found in most documentation.
But I think we could live with a specific SQL loading method for the 10%.

I personally would appreciate to have CREATE EXTENSION postgis; (or
postgis_core) and CREATE EXTENSION postgis_raster;
As said, it is then clear from an application point of view which
dependencies you have, avoid to rely on GDAL if not needed - which can
be a pain with all its dependencies - and avoid to clutter the function
lists, memory and all.

But then it is a matter of how much work is required for that from the
developers, and is there funding to make it work this way.

Vincent

On 27/10/2015 23:49, Paragon Corporation wrote:
> How many folks are unable to compile postgis with raster or have a
> package that has no raster support and who REALLY REALLY want to be
> able to install postgis with:
> 
> CREATE EXTENSION postgis;
> 
> This is a repost of a conversation I started in PostGIS -develop
> which I probably should have cc'd PostGIS users on.  It seems not too
> many people on dev have issue compiling with GDAL support.
> 
> The issue discussed is that since postgis raster is part of the
> postgis extension, if people choose to compile without raster support
> because they can't get GDAL to work (despite being discouraged), they
> can not do
> 
> CREATE EXTENSION postgis;
> 
> They have to resort to the old way of using postgis contrib scripts
> to install.
> 
> My proposition https://trac.osgeo.org/postgis/ticket/3338
> 
> 
> was still have a postgis that has raster support (as breaking it out
> at this point would be  headache particularly for the person who has
> to deal with the upgrade ramifications) and a big headache for people
> who use raster heavily as now they'd have two extensions to bother
> with.
> 
> That said the proposition in a nutshell is this
> 
> 1) Still have raster as default and you still need to explicitly ask
> for --without-raster 2) If you ask for --without-raster, you'll get
> an extension called  "postgis" with no raster functions in it, but
> with an extension version number that reads  2.3.0-no-raster
> 
> So a query like this if you compiled without raster:
> 
> SELECT  * FROM pg_available_extensions() WHERE name = 'postgis';
> 
> Would read:
> 
> name   | default_version |                               comment 
> ---------+-----------------+---------------------------------------------------------------------
>
> 
postgis | 2.3.0-no-raster     | PostGIS geometry and geography  spatial
types and functions
> (1 row)
> 
> Instead of:
> 
> name   | default_version |                               comment 
> ---------+-----------------+---------------------------------------------------------------------
>
> 
postgis | 2.3.0           | PostGIS geometry, geography, and raster
spatial types and functions
> (1 row)
> 
> 
> Which means people who have postgis without raster support can follow
> the world renowned instructions:
> 
> CREATE EXTENSION postgis;
> 
> Like everyone else and all the other postgis extensions
> postgis_topology, postgis_tiger_geocoder will be built as well.
> 
> The main downside to this is 1) you have a postgresql postgis
> extension that may not have any raster functions and one that goes by
> same extension name postgis with raster support. Though I claim this
> is clear from the version number and description :) 2) It does mean
> it’s a bit easier for packager maintainers to package a postgis
> without raster support which is both good and bad -- good for older
> systems where an old antiquated gdal is not readily available so they
> can still package a newer postgis for it. -- bad for users who rely
> on package maintainers and really want raster support as their
> package maintainer might just decide not to bother with raster.
> 
> If there is enough interest, I will reopen the ticket I closed and
> create and RFC as Paul Ramsey suggested to flesh out the complete
> details of this and how upgrading would work For folks coming from a
> postgis without raster non-extension version to a postgis without
> raster extension version.
> 
> Thanks, Regina http://www.postgis.us http://postgis.net
> 
> 
> 
> 
> -----Original Message----- From: postgis-devel
> [mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Sandro
> Santilli Sent: Tuesday, October 27, 2015 5:28 PM To: PostGIS
> Development Discussion <postgis-devel at lists.osgeo.org> Subject: Re:
> [postgis-devel] postgis extension sans raster (only for folks who
> can't compile with raster support) - PSC Vote and developer/packager
> comments please
> 
> On Tue, Oct 27, 2015 at 03:39:29PM -0400, Paragon Corporation wrote:
>> I've retracted.  Seems not enough interest for this and could lead
>> to confusion.
>> 
>> Or maybe I should have asked on postgis-users :)
> 
> That's a good idea. Ask on postgis-users if there's any interest
> about being able to install postgis with no raster support.
> 
> --strk; _______________________________________________ postgis-devel
> mailing list postgis-devel at lists.osgeo.org 
> http://lists.osgeo.org/mailman/listinfo/postgis-devel
> 
> 
> _______________________________________________ postgis-devel mailing
> list postgis-devel at lists.osgeo.org 
> http://lists.osgeo.org/mailman/listinfo/postgis-devel
>