Implement Uri.resolve(Uri)

Implements method Uri.resolve(Uri). This has two benefits:

1. More efficient than the existing implementation. Using the Uri test class's resolve Uris, this method performed 30-45% better than existing.

2. Removes java.net.URI from the resolve(...) code path. This makes it possible for Uris parsed leniently by an overridden UriParser to be resolvable.

[awiner, 2009-03-26 15:50:55]

Some comments about performance.

http://codereview.appspot.com/27117/diff/1/2
File java/common/src/main/java/org/apache/shindig/common/uri/Uri.java (right):

http://codereview.appspot.com/27117/diff/1/2#newcode142
Line 142: } else if (authority != null) {
This might be clearer if there was progressive nesting, so:

} else {
  scheme = getScheme();
  if (authority == null) {
    authority = getAuthority();
    if (... )
  }
etc...

Then again, it might not.

http://codereview.appspot.com/27117/diff/1/2#newcode184
Line 184: Stack<String> pathStack = new Stack<String>();
java.util.Stack is largely deprecated.  (It extends *Vector*, yikes.)

http://codereview.appspot.com/27117/diff/1/2#newcode186
Line 186: Collections.addAll(pathStack, curPath.split("/"));
I think it's faster to use StringTokenizer:  String.split() parses a regex. 
Plus, this builds an array just to throw it away.

http://codereview.appspot.com/27117/diff/1/2#newcode193
Line 193: LinkedList<String> mergePath = new LinkedList<String>();
ArrayList would be faster

http://codereview.appspot.com/27117/diff/1/2#newcode194
Line 194: Collections.addAll(mergePath, otherPath.split("/"));
split() -> StringTokenizer, ditto

[plindner, 2009-03-26 16:01:14]

It might be nice to have some tests.  java.net.Uri has the following:

http://www.google.com/codesearch/p?hl=en#TTY8xLpnKOE/test/java/net/URI/Test.j...

http://codereview.appspot.com/27117/diff/1/2
File java/common/src/main/java/org/apache/shindig/common/uri/Uri.java (right):

http://codereview.appspot.com/27117/diff/1/2#newcode194
Line 194: Collections.addAll(mergePath, otherPath.split("/"));
Suggest using StringUtils.splitPreserveAllTokens or StringUtils.split
(non-regex).

I *think* preserveAllTokens will avoid the if statement below...