[PROJ] Am I supposed to get same results with libproj if passing proj4 +proj from/to crs ?

[Even Rouault]

Sandro,

there is a NAD27 to WGS84 datum shift involved. So this might depend on 
the transformations registered in the PROJ database of the version you 
use and the grids available.

Result (-11708740.522541871, 3209723.515103944) is suboptimal as it is 
the you get when no grids are used

Result (-11708734.486616733, 3209717.223316072) is the one using the 
conus / us_noaa_conus.tif grid

Whether PROJ networking is enabled (PROJ_NETWORK=ON) can influence the 
result (if it is set, then the conus grid will be used even if not 
installed locally)

Even

Le 10/11/2021 à 00:02, Sandro Santilli a écrit :
> I'm trying to obtain the *same* transformations from different
> libproj versions (5.2.0 and 7.2.1) by passing the *same*
> proj4 strings to the transformation, but I'm instead getting
> different results.
>
> In order to more easily test this I'm a small python script
> using pyproj:
>
>    http://strk.kbt.io/tmp/transform_test.py
>
> The script projects coordinates 0,0 from EPSG:32040
> to EPSG:3857, using the proj4 string found in
> PostGIS 3.1.4 spatial_ref_sys for those SRIDs.
>
> On a machine (pyproj-3.2.1 using libproj-ce980207.so.22.1.1)
> I'm getting these coordinates in output:
>
>    (-11708740.522541871, 3209723.515103944)
>
> On another machine (pyproj-2.6.1 using libproj.so.19)
> I'm getting these coordinates:
>
>    (-11708734.486616733, 3209717.223316072)
>
> Can you think of a reason why I'd get different numbers ?
> Would it be a regression for the libproj-ce980207.so.22.1.1
> version to return different numbers ?
>
> Please keep me in Cc in case of replies as I don't think I'm
> receiving emails sent to this list.
>
> --strk;
>
>    Libre GIS consultant/developer
>    https://strk.kbt.io/services.html
> _______________________________________________
> PROJ mailing list
> PROJ at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/proj

-- 
http://www.spatialys.com
My software is free, but my time generally not.