Fix crash in close_ps() for NaN input coordinates.

The Assert() here seems unreasonably optimistic.  Andreas Seltenreich
found that it could fail with NaNs in the input geometries, and it
seems likely to me that it might fail in corner cases due to roundoff
error, even for ordinary input values.  As a band-aid, make the function
return SQL NULL instead of crashing.

Report: <87d1md1xji.fsf@credativ.de>

diff --git a/src/backend/utils/adt/geo_ops.c b/src/backend/utils/adt/geo_ops.c
index 77871b10ffbe2e41fb7c22dc9e20c9adf86d4a44..ee60203f998c0ca5fcaefb6d6a7de063a0f4f1ac 100644 (file)
--- a/src/backend/utils/adt/geo_ops.c
+++ b/src/backend/utils/adt/geo_ops.c
@@ -2877,7 +2877,7 @@ close_ps(PG_FUNCTION_ARGS)
    }
 
    /*
-    * at this point the "normal" from point will hit lseg. The closet point
+    * at this point the "normal" from point will hit lseg. The closest point
     * will be somewhere on the lseg
     */
    tmp = line_construct_pm(pt, invm);
@@ -2886,7 +2886,15 @@ close_ps(PG_FUNCTION_ARGS)
           tmp->A, tmp->B, tmp->C);
 #endif
    result = interpt_sl(lseg, tmp);
-   Assert(result != NULL);
+
+   /*
+    * ordinarily we should always find an intersection point, but that could
+    * fail in the presence of NaN coordinates, and perhaps even from simple
+    * roundoff issues.  Return a SQL NULL if so.
+    */
+   if (result == NULL)
+       PG_RETURN_NULL();
+
 #ifdef GEODEBUG
    printf("close_ps- result.x %f  result.y %f\n", result->x, result->y);
 #endif