[postgis-devel] Questions about compliance of PostGIS with OGC standards

[Paul Ramsey]

> On Jan 5, 2021, at 1:32 PM, Aleksandra Stasiak <aleksandra.stasiak at wmii.uni.lodz.pl> wrote:
> 
> Dear Sir/ Madam
> I am an academic teacher and I am trying to write a scientific article about some aspects of databases and their extesions and I have some problems with understanding a few things connected with PostGIS (and compliance with the OGC standards). Could I ask you for help?
> 
> There is a sentence in the PostGIS documentation saying that "PostGIS is compliant with the Open Geospatial Consortium's (OGC) OpenGIS Specifications". However, on OGC websites, when we search for products compliant with the Simple Feature Access standard or with an older version, i.e. Simple Features SQL, PostGIS does not appear if we select products with the "Compliant Products Only" option or when we select products with the "Reference Implementations Only" option, only by searching with the "All Implementations" option. Why? Why is it not just compliant with these specifications? What kind of implementation of standards is that?

https://www.ogc.org/resource/products/compliant?display_opt=3&specid=522
https://www.ogc.org/resource/products/compliant?display_opt=3&specid=149

I see Crunchy Certified PostGIS in there a couple times. Also old submissions by Refractions and Boundless.
In any event, OGC Certification is a pay-to-play deal. We are "compliant" because we've followed and implemented the standard. The entry in the cerfified list are because Crunchy has spent the time to run the tests and submit them to OGC, and paid OGC some money to be listed as "officially certified". 

In a very "shrinkwrapped software" touch, OGC also ties certifications to software release version, so in theory we have to go back and re-cerfity each time we do a release. As you can imagine, we don't have much appetite for that.

> I have also noticed that PostGIS has the functions converting from WKT format such as ST_PointFromText, ST_LineFromText, ST_PolygonFromText, ST_GeomCollFromText, ST_MPointFromText etc. (for all clasess of geometries) with a more general ST_GeomFromText function. But for the WKB format there is no function like  ST_PolygonFromWKB or ST_GeomCollFromWKB, there are only ST_PointFromWKB, ST_LineFromWKB (or ST_LinestringFromWKB as alias) and of course more general ST_GeomFromWKB. Why PostGIS does not have functions converting from WKB for all classes (as it does for WKT)? It is not a problem? ST_GeomFromWKB is enough for compliance?

If you read the actual compliance test, you'll find it's a joke. It was never updated for standards after the 1.0 standard, and is internally inconsistent in a number of places. In implementing SQL/MM, which is where a lot of these duplicative function signatures come from, we eventually stopped. For our implementation, which has just one actual database type (geometry) having 7 separate constructors (14! both WKT and WKB!) is pretty silly, and we probably shouldn't have done the WKT ones, but once they are added it's hard to remove them.

> In many places of the documentation there are fragments saying that some function "Return a specified ST_Geometry value" (for instance in description of function ST_GeometryFromText). I know that ST_GEOMETRY is the root of the SQL/MM hierarchy of classes, but I don't understand these descriptions - PostGIS doesn't technically have ST_GEOMETRY type (it has geometry, geography, and some types for boxes). So does ST_GEOMETRY value denote here a value with geometry type, with geometry that would belong to the ST_GEOMETRY class according to the SQL/MM classification? 

Yes, it looks like the documents use ST_Geometry and Geometry interchangeably. The only place where ST_ or not has a practical effect is the ST_GeometryType() and GeometryType() functions, that actually return different answers.


> Is the compliance with OGC standard concern geometry type or geography type (or both)?

OGC has nothing to say about geography, and has published no standards about how to handle geodetic objects. It's a massive blind spot but I guess none of their sponsors really cares. So for compliance only geometry really "matters", even though the function names are common between the two of them for a lot of things.

ATB,

P



> I would be grateful for explaining these things to me.
> Best wishes,
> Aleksandra Stasiak
> <Outlook-0k0ma5sf.png>
> This message and its attachments are confidential and protected by law. If you are not the intended recipient of the message, please contact the  sender immediately and delete the message with its attachments. Also, in that case, further dissemination of the message or its  attachments, as well as sharing or disclosing their content in its  entirety or part is strictly forbidden.
> 
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/postgis-devel