[postgis-devel] PSC Vote: Make spatial_ref_sys as a view

[Sandro Santilli]

On Sat, Feb 05, 2022 at 02:42:27AM +0300, Darafei "Komяpa" Praliaskouski wrote:
> I believe spatial_ref_sys by itself is misdesign from the same spec that
> holds geometries as tables of x-y's with order number, similar to
> geometry_columns - useful in databases that don't have spatial types or
> observability of schema.
> 
> We have a problem supporting coordinate epochs that does not nicely fit the
> spatial_ref_sys design.
> 
> +0 to wiggle the code in this area to have some attention to it. I
> believe real profits will come from redesigning projection handling so that
> spatial_ref_sys is obsolete.

I also think a redesign is useful, but the ability to "tag" Geometry
values with a small integer as a foreign key to a table holding more
information about some aspects of it (this is different from what
geometry_columns was) seems still useful to me, so while I see the table
possibly changing in the future, I don't forsee a complete drop of it.

--strk;