pg_upgrade fails due to lack of dependencies.

Regina Obe lr at pcorp.us
Mon Jun 30 10:50:37 PDT 2025


> 
> > On Jun 30, 2025, at 10:44 AM, Regina Obe <lr at pcorp.us> wrote:
> >
> > I’m still confused how having a dependency between spatial_ref_sys and
> geography would guarantee that the spatial_ref_sys data will get loaded
> before anything that references it is created.
> >  Ultimately you’ll run into the issue spatial_ref_sys doesn’t have record for
> 4326 or 3857.
> >  To fix the issue that jeevanchalke had raised, I think Paul decided the cleaner
> way is just to get rid of the longlat validation.  But that won’t fix your issue
> because you are doing an ST_Transform.
> >  Like I said a dependency on a structure just guarantees the structure has to
> exist.
> 
> So your concern is that even with a dep between st_transform and
> spatial_ref_sys, it’s entirely possible that pg_restore will just restore the empty
> table, and the problem will still occur?
> 

YES.





More information about the postgis-devel mailing list