From: ellson <devnull@localhost>
Date: Thu, 15 Dec 2005 01:57:22 +0000 (+0000)
Subject: Workaround for a problem in gcc-4.1.0 when used with -O2
X-Git-Tag: LAST_LIBGRAPH~32^2~6922
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=a113d70fee9acb743c08a352672d0f94603130e9;p=graphviz

Workaround for a problem in gcc-4.1.0 when used with -O2
---

diff --git a/lib/pathplan/route.c b/lib/pathplan/route.c
index 80892fdb6..cdfc6411f 100644
--- a/lib/pathplan/route.c
+++ b/lib/pathplan/route.c
@@ -262,7 +262,21 @@ static int mkspline(Ppoint_t * inps, int inpn, tna_t * tnas, Ppoint_t ev0,
 	x[0] += dot(tnas[i].a[0], tmp);
 	x[1] += dot(tnas[i].a[1], tmp);
     }
+#if 0
     det01 = c[0][0] * c[1][1] - c[1][0] * c[0][1];
+#else
+    /* workaround for problem with:
+     *         gcc (GCC) 4.1.0 20051212 (Red Hat 4.1.0-0.7)
+     *
+     * the problem is a NAN from the subtraction which 
+     * shows up as a SIGFPE a few lines down when
+     * det01 is used as the denominator in a division 
+     */
+    det01 = c[0][0] * c[1][1];
+    d01 = c[1][0] * c[0][1];
+    if (ABS(d01) > 1e-6)
+	det01 -= d01;
+#endif
     det0X = c[0][0] * x[1] - c[0][1] * x[0];
     detX1 = x[0] * c[1][1] - x[1] * c[0][1];
     if (ABS(det01) >= 1e-6) {