Even Rouault

Slightly related to that topic : https://github.com/OSGeo/PROJ/pull/2024 . This will recognize "official" WKT1:ESRI syntax for VERTCS[]. Will not directly help for the 2 cases mentioned in https://github.com/OSGeo/PROJ/issues/2022#issue-576210614

> WKID crs values. WKID is a concept that only works when you have the equivalent of a spatial_ref_sys column to point to. Often people make the confusion WKID =...

> do you see major downsides in allowing authority/code pairs ? that's a convenient way of communicating a CRS when all parties have the database available. Orthogonal to that issue,...

> since WKB specifies (x, y). I'm going to be too pedantic, but 'x' and 'y' doesn't mean a lot. If you look at the name of the axis of...

> here the Srid is in the header itself. Warning: as I pointed in https://github.com/geopandas/geo-arrow-spec/issues/8#issuecomment-1026800848, a SRID is a pointer to an entry in a spatial_ref_sys table. SRID != EPSG...

> Are WKT2 strings guaranteed to stay the same for a given EPSG code, say for `4326`? no, that one will typically change over time as new members of the...

A better string based approach to identify EPSG:4326 would be to check that the string ends with (*not* just contains) ``ID["EPSG",4326]]`` (however one cannot exclude that the formatting will be...

> we have some "crs information" from the shapefile but with no guarantee that it's formatted as WKT2 (I think). The WKT in a Shapefile .prj is *never* (the non-deprecated...

> As far as I can tell, the `geo` metadata is at the file level, not the `RecordBatch` level. well, at least the arrow::ipc::RecordBatchFileReader::metadata() (" Return the contents of the...

> I suppose that if we wanted to make use of this, that could be changed (or an API could be added to provide the metadata when closing the file)....