[PROJ] Problems with transformations: projection could not be created and swapped axes with PJ_TYPE_VERTICAL_CRS

[support at mnspoint.com]

Some answers:
------------------------------------------------------
Sebastian Gellrich via PROJ kirjoitti 2023-11-03 17:32:
> Hello PROJ9 Community,
> 
> I have 2 questions about transformations. First of all, I do not know
> one hundred percent about transformations.
> 
> I use a library I built myself for all possible Android platforms.
> 
> Now I am testing the library a little and have encountered the
> following problems:
> 
> 1.
> 
> I get a list of all CRS (PJ_TYPE_CRS) with
> proj_get_crs_info_list_from_database(...).
> 
> For testing purposes, I then create a projection with all CRSs using
> proj_create_crs_to_crs(...), which I then normalize with
> 
> proj_normalize_for_visualization(...).
> 
> My source is always WGS84 (EPSG:4326).
> 
> With some projections I get the error "No inverse operation"
> (PROJ_ERR_OTHER_NO_INVERSE_OP) after calling
> proj_get_crs_info_list_from_database.
> 
> What does this mean? Can I somehow create this projection differently?
> 
> 
> Example: EPSG:2218
> 

"No inverse operation" means that the projection is just defined 
forward. So, you cannot use that projection inverse and the library does 
not have any way to calculate inverse. The inverse is not defined. So, 
most likely you should remove all such projections from your procedure 
or list?

> With some CRS of type PJ_TYPE_VERTICAL_CRS I get the error code
> "Unknown error (code 4096)".
> 
> Here the error code tells me even less what the problem is.
> 
> Examples:           EPSG:5620
> 
>                                EPSG:6357 (here I also received the log
> message "pipeline: Pipeline: Mismatched units between step 3 and 4"
> from Proj9)
> 

I don't know what that is. Might be related to previous or similar.

> 2.
> 
> When transforming to type PJ_TYPE_VERTICAL_CRS, the axes seem to be
> swapped in the result.

That is much possible in many cases. I assume you just have to live with 
it or do some fix in your own coding?!

> 
> I used the same WGS84 coordinate for all tests (x=14, y=52, z=100 =>
> Germany, Brandenburg).
> 
> The transformed coordinates were all x=52 and y=14. Is this
> intentional?
> 
> Some transformed coordinates also contained only Infinity.
> 
> Thanks in advance for your help.
> _______________________________________________
> PROJ mailing list
> PROJ at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/proj

Regards: Janne