why are p#d_same using FP_EQUALS ?

[Paul Ramsey]

I do not remember, but there can be no doubt that there are cases that depend on them, so changing them will result in regressions. Since it all narrows down to a fine point in those function definitions, an obvious first test is to redefine them to exact tests and see what breaks.

P

> On Mar 22, 2024, at 10:10 AM, Sandro Santilli <strk at kbt.io> wrote:
> 
> Given it's the second time that I stumble upon issues related
> to use of FP_EQUALS in the p2d_same and p4d_same functions I'm
> wondering why those FP_EQUALS macros were added at all.
> 
> Does anyone remember the rationale for that ?
> 
> The two cases:
>  - https://trac.osgeo.org/postgis/ticket/5698
>  - https://trac.osgeo.org/postgis/ticket/5394
> 
> In topology I've stopped using the p2d_same method for the
> problem using tolerance raises, but this time the problem
> was in the ST_Split function itself. Same cause (considering
> two almost-but-not-exactly-equal points equal).
> 
> Should we stop that ?
> 
> --strk; 
> 
>  Libre GIS consultant/developer
>  https://strk.kbt.io/services.html