Description:
------------
When dumping a float number, it is represented incorrectly. Generally, in some cases, PHP handles float numbers in invalid format.

It is related to a setting of php.ini directive "precision". When a float number with one decimal place is dumped and high precision is set, it is not dumped in an original format, but in damaged one.

Additional info:
32-bit system
CPU: AMD Athlon64

Reproduce code:
---------------
foreach (range(0, 30) as $precision) {
	ini_set('precision', $precision);
	echo $precision . ': ';
	var_dump(20.2);
}

Expected result:
----------------
0: float(20)
1: float(20.2)
2: float(20.2)
3: float(20.2)
4: float(20.2)
5: float(20.2)
6: float(20.2)
7: float(20.2)
8: float(20.2)
9: float(20.2)
10: float(20.2)
11: float(20.2)
12: float(20.2)
13: float(20.2)
14: float(20.2)
15: float(20.2)
16: float(20.2)
17: float(20.2)
18: float(20.2)
19: float(20.2)
20: float(20.2)
21: float(20.2)
22: float(20.2)
23: float(20.2)
24: float(20.2)
25: float(20.2)
26: float(20.2)
27: float(20.2)
28: float(20.2)
29: float(20.2)
30: float(20.2)

Actual result:
--------------
0: float(2.0E+1)
1: float(2.0E+1)
2: float(20)
3: float(20.2)
4: float(20.2)
5: float(20.2)
6: float(20.2)
7: float(20.2)
8: float(20.2)
9: float(20.2)
10: float(20.2)
11: float(20.2)
12: float(20.2)
13: float(20.2)
14: float(20.2)
15: float(20.2)
16: float(20.2)
17: float(20.199999999999999)
18: float(20.1999999999999993)
19: float(20.19999999999999929)
20: float(20.199999999999999289)
21: float(20.1999999999999992895)
22: float(20.19999999999999928946)
23: float(20.199999999999999289457)
24: float(20.1999999999999992894573)
25: float(20.19999999999999928945726)
26: float(20.199999999999999289457264)
27: float(20.1999999999999992894572642)
28: float(20.19999999999999928945726424)
29: float(20.19999999999999928945726424)
30: float(20.1999999999999992894572642399)