[postgis-devel] Proj 9.2

[Paul Ramsey]

> On Jun 19, 2023, at 12:02 PM, Even Rouault <even.rouault at spatialys.com> wrote:
> 
> Paul,
>> I was wondering why we were only seeing new proj regressions with proj 9.2 and I think it is this change in policy that is driving it. From proj.ini:
>> 
>> ; Can be set to on so that by default the lack of a known resource files needed
>> ; for the best transformation PROJ would normally use causes an error, or off
>> ; to accept missing resource files without errors or warnings.
>> ; This default value itself is overriden by the PROJ_ONLY_BEST_DEFAULT environment
>> ; variable if set, and then by the ONLY_BEST setting that can be
>> ; passed to the proj_create_crs_to_crs() method, or with the --only-best
>> ; option of the cs2cs program.
>> ; (added in PROJ 9.2)
>> ; only_best_default = on
>> 
>> Basically until recently proj was quietly falling back to less accurate transforms when grids were missing. Now the default policy is to complain and fail if the best transform is missing.
> 
> Actually no. The default is only_best=no and still to not complain and fail, but to implement the possibility of setting only_best=yes, there were non trivial code changes which turned to cause behaviour changes (including transformations to fail whereas they should have used dummy/ballpark transformations) that we noticed post 9.2.0 and hopefully most of which have been resolved with 9.2.1.

Ah, sorry my mistake, I was using my Postgres goggles in reading the config file: Pg puts the default value into the commented out variable, which makes finding the defaults pretty easy, and avoids any accidental changes in policy (since a change requires both an uncommenting and a value changing). 

> On which precise transformations do you see changes?

There were quite a few in our regression tests but we’ve been working on wiping them out so we can run tests in isolation without hitting the grid shift code line at all, I’m afraid I don’t have them at hand anymore

P

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