This test is failing because of the good old-well known floating point
drift issue.
4.9 + 0.1 > 5.0
The kludge attached below "solves" it, but... well... it's a kludge. It
basicly adds (or subtracts) the smallest measurable piece of a double
during the comparison step (so as to not alter the actual value) and give
the step some room to breathe.
Isn't there a better way to do it?
Index: ext/standard/array.c
===================================================================
RCS file: /repository/php4/ext/standard/array.c,v
retrieving revision 1.224
diff -u -r1.224 array.c
--- ext/standard/array.c 1 Apr 2003 21:47:21 -0000 1.224
+++ ext/standard/array.c 2 Apr 2003 01:17:29 -0000
@@ -81,6 +81,8 @@
#define INTERSECT_NORMAL 0
#define INTERSECT_ASSOC 1
+#define DOUBLE_DRIFT_FIX 0.000000000000001
+
PHP_MINIT_FUNCTION(array)
{
#ifdef ZTS
@@ -1543,7 +1545,7 @@
err = 1;
goto err;
}
- for (; low >= high; low -= step) {
+ for (; low >= (high - DOUBLE_DRIFT_FIX); low -=
step) {
add_next_index_double(return_value, low);
}
} else if (high > low) { /* Positive steps */
@@ -1551,7 +1553,7 @@
err = 1;
goto err;
}
- for (; low <= high; low += step) {
+ for (; low <= (high + DOUBLE_DRIFT_FIX); low +=
step) {
add_next_index_double(return_value, low);
}
} else {
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
