pgsql: Tweak "line" test to avoid negative zeros on some platforms

Peter Eisentraut <peter_e@gmx.net> writes:

Tweak "line" test to avoid negative zeros on some platforms

I had been playing around with that last night on a PPC-era Mac,
hoping to get rid of the problem by twiddling the expressions in
line_construct_pts(). The only thing I could find to fix it
was the brute-force approach:

diff --git a/src/backend/utils/adt/geo_ops.c b/src/backend/utils/adt/geo_ops.c
index 25f0bfd..41178a6 100644
--- a/src/backend/utils/adt/geo_ops.c
+++ b/src/backend/utils/adt/geo_ops.c
@@ -1116,6 +1116,9 @@ line_construct_pts(LINE *line, Point *pt1, Point *pt2)
 		line->A = (pt2->y - pt1->y) / (pt2->x - pt1->x);
 		line->B = -1.0;
 		line->C = pt1->y - line->A * pt1->x;
+		/* on some platforms, the preceding expression tends to produce -0 */
+		if (line->C == 0.0)
+			line->C = 0.0;
 #ifdef GEODEBUG
 		printf("line_construct_pts- line is neither vertical nor horizontal (diffs x=%.*g, y=%.*g\n",
 			   DBL_DIG, (pt2->x - pt1->x), DBL_DIG, (pt2->y - pt1->y));

While this is surely pretty ugly, I think it is less ugly and more robust
than what you've done here. Comments?

regards, tom lane

--
Sent via pgsql-committers mailing list (pgsql-committers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-committers

On 10/25/13, 10:55 AM, Tom Lane wrote:

diff --git a/src/backend/utils/adt/geo_ops.c b/src/backend/utils/adt/geo_ops.c
index 25f0bfd..41178a6 100644
--- a/src/backend/utils/adt/geo_ops.c
+++ b/src/backend/utils/adt/geo_ops.c
@@ -1116,6 +1116,9 @@ line_construct_pts(LINE *line, Point *pt1, Point *pt2)
line->A = (pt2->y - pt1->y) / (pt2->x - pt1->x);
line->B = -1.0;
line->C = pt1->y - line->A * pt1->x;
+		/* on some platforms, the preceding expression tends to produce -0 */
+		if (line->C == 0.0)
+			line->C = 0.0;
#ifdef GEODEBUG
printf("line_construct_pts- line is neither vertical nor horizontal (diffs x=%.*g, y=%.*g\n",
DBL_DIG, (pt2->x - pt1->x), DBL_DIG, (pt2->y - pt1->y));

While this is surely pretty ugly, I think it is less ugly and more robust
than what you've done here. Comments?

Hehe, if that work's, then yes, it's better.

--
Sent via pgsql-committers mailing list (pgsql-committers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-committers