Fixing analyzer warnings on overloading == but not get hashCode.

Fixing analyzer warnings on overloading == but not get hashCode.
BUG=11613
R=sra@google.com

Committed: https://code.google.com/p/dart/source/detail?r=24614

[sra1, 2013-06-28 21:37:03]

lgtm with a better hash function.

https://chromiumcodereview.appspot.com/18034021/diff/4001/sdk/lib/html/dart2j...
File sdk/lib/html/dart2js/html_dart2js.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/4001/sdk/lib/html/dart2j...
sdk/lib/html/dart2js/html_dart2js.dart:24513: 
I suggest making a good 4 input hash function modelled after the string hash
function in sdk/lib/_internal/lib/js_string.dart

You don't have to combine the hashCodes of the components if they are already
ints (but do if they may be null or num).

int get hashCode => JenkinsSmiHash.hash4(left, top, width, height);
--or--
int get hashCode => JenkinsSmiHash.combine4(left.hashCode, top.hashCode,
width.hashCode, height.hashCode);

class JenkinsSmiHash {
  static int combine(int hash, int value) {
    // inner loop code
  }

  static int finish(int hash) {
    // wrap up.
  }

  static int combine2(int hash, int a, int b) =>  combine(combine(h, a), b);
  static int combine4(int hash, int a, int b, int c, int d) =>
      combine2(combine(h, a, b), c, d);

  int hash2(int a, int b) => finish(combine2(0, a, b));

  int hash4(int a, int b, int c, int d) => finish(combine4(0,a,b,c,d));

}


You can probably get away with just combining if the inputs are already well
hashed (a change in one bit in the input causes approx half the bits in the
hashCode to flip)

[sra1, 2013-06-28 22:25:02]

lgtm, the test could be improved if you like.

https://chromiumcodereview.appspot.com/18034021/diff/9001/sdk/lib/html/html_c...
File sdk/lib/html/html_common/jenkins_smi_hash.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/sdk/lib/html/html_c...
sdk/lib/html/html_common/jenkins_smi_hash.dart:38: static int hash2(a, b) =>
JenkinsSmiHash.finish(
Do you need the class name here?

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/point_te...
File tests/html/point_test.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/point_te...
tests/html/point_test.dart:112: var c = new Point(1, 1);
A fun test is that (0,1) and (1,0) differ.

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/rect_tes...
File tests/html/rect_test.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/rect_tes...
tests/html/rect_test.dart:156: var c = new Rect(1, 1, 2, 3);
Rect(1,0,2,3) Rect(0,1,3,2) might be a better tests since they would have been
the same with xor.

[blois, 2013-06-28 22:35:37]

https://chromiumcodereview.appspot.com/18034021/diff/9001/sdk/lib/html/html_c...
File sdk/lib/html/html_common/jenkins_smi_hash.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/sdk/lib/html/html_c...
sdk/lib/html/html_common/jenkins_smi_hash.dart:38: static int hash2(a, b) =>
JenkinsSmiHash.finish(
On 2013/06/28 22:25:02, sra1 wrote:
> Do you need the class name here?

Done.

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/point_te...
File tests/html/point_test.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/point_te...
tests/html/point_test.dart:112: var c = new Point(1, 1);
On 2013/06/28 22:25:02, sra1 wrote:
> A fun test is that (0,1) and (1,0) differ.

Done.

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/rect_tes...
File tests/html/rect_test.dart (right):

https://chromiumcodereview.appspot.com/18034021/diff/9001/tests/html/rect_tes...
tests/html/rect_test.dart:156: var c = new Rect(1, 1, 2, 3);
On 2013/06/28 22:25:02, sra1 wrote:
> Rect(1,0,2,3) Rect(0,1,3,2) might be a better tests since they would have been
> the same with xor.

Done.

[blois, 2013-06-28 22:36:58]

Committed patchset #4 manually as r24614 (presubmit successful).