[postgis-devel] Proj GUCs

[Even Rouault]

If for your tests you want to have the greater stability for the results 
of coordinate transformations, you should try to stay in the same datum 
in the source and target CRS. That is projecting between different 
projections of WGS 84 (or NAD83 or whatever) should give you pretty 
stable results (unless the maths behind a projection method are changed, 
but that's much less frequent than datum transformation parameters). But 
things like NAD83 <--> WGS 84 might vary over PROJ versions and depend 
on grids.

Even

Le 21/06/2023 à 17:02, Regina Obe a écrit :
> it, that's fine -- but it's not really a postgis issue.
>> I would say that postgis tests should expect full-precision/best-choice
>> transforms and that the tests should say that someplace.  I am unclear on
>> this for many reasons, but the top one is that I have not yet updated proj
> in
>> pkgsrc to current, partly because of not getting to it, but mostly because
> proj
>> updates the last few years have been leaving unmaintained and
>> undermaintained packages behind.
> I think the regressions we were having have been largely resolved.
> The main regression was for our proj cache overflow tests, which doesn't
> even care about the answers
> but more that the timing isn't horrible.
>
> But I think for that we have now settled on 100 someodd srids that don't
> break on proj 4.9 - 9.2
> regardless what the proj settings are.
>
> Most of the ones where we check answers I think are working fine cause they
> never relied on grid shifts.
>
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/postgis-devel

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