Fwd: [Geopackage] Is mod_spatialite recommended?

[Even Rouault]

CC'ing an exchange from the geopackage mailing list about a slight non conformance from Spatialite GeoPackage implementation

-------- Message transféré --------

Sujet :	Re: [Geopackage] Is mod_spatialite recommended?
Date :	Wed, 26 Jun 2024 14:45:48 +0000
De :	Jeff Yutzler via Geopackage <geopa...@lists.ogc.org>
Répondre à :	Jeff Yutzler <je...@imagemattersllc.com>
Pour :	geopa...@lists.ogc.org <geopa...@lists.ogc.org>, Thomas Larsen Wessel <mrv...@gmail.com>

Thomas,

I can't speak to Spatialite other than to say that GeoPackage was derived from it, not the other way around. 

It would not be unreasonable for us to relax Requirement 11, or at least the abstract test thereof, to allow a case-insensitive description. This is quite pedantic as test failures go, so I wrote a ticket. The establishment of Requirement 11 happened around 2013, before I became SWG chair. I did not participate in the discussions that established it and have no idea why we depart from what Spatialite was doing originally.

Meanwhile, we have a list of known software implementations here.

-Jeff

---

From: Geopackage <geopackage-bounces+jeffy=imagematt...@lists.ogc.org> on behalf of Thomas Larsen Wessel via Geopackage <geopa...@lists.ogc.org>
Sent: Tuesday, June 25, 2024 4:48 PM
To: geopa...@lists.ogc.org <geopa...@lists.ogc.org>
Subject: [Geopackage] Is mod_spatialite recommended?

 

You don't often get email from geopa...@lists.ogc.org. Learn why this is important

I was surprised to see that the GeoPackage file I created with the gpkg* functions in Spatialite (version 5.0.1) does not pass the official conformance test suite (https://cite.opengeospatial.org/teamengine/). 

Lets consider the simplest example. Spatialite's gpkgCreateBaseTables creates gpkg_spatial_ref_sys with a few default rows. SpatiaLite uses the string "Undefined", where the test suite expects "undefined", and thus it fails requirement 11. 

That would be really simple to fix, yet nobody fixed it. That kind of suggests lack of ambition, or at least a lack of resources. Or have I misunderstood something? Is the test suite misleading somehow?

When even the simplest to implement requirements are not met, I doubt I can trust that the file won't also be broken in other ways. 

Should I look for a different tool than SpatiaLite? Are there any other extensions for sqlite3 that can help build a GeoPackage?

Sincerely, Thomas