Summary

Update the java.net.URI::compareTo specification and behavior to adhere to the strong recommendation of the Comparable interface, that (x.compareTo(y)==0) == (x.equals(y)), for URIs that contain hexadecimal digits of escaped octets that differ only in case.

Problem

The URI::compareTo method does not strictly follow the strong recommendation of Comparable::compareTo. Specifically, URI::compareTo does not consider two URIs to be equal when they differ only in the case of hexadecimal digits of escaped octets. While URI::equals does consider such URIs to be equal.

For example:

jshell> var uri1  = URI.create("http://example.com/%5bsegment%5d")
uri1 ==> http://example.com/%5bsegment%5d

jshell> var uri2  = URI.create("http://example.com/%5Bsegment%5D")
uri2 ==> http://example.com/%5Bsegment%5D

jshell> uri1.equals(uri2)
$4 ==> true

jshell> uri1.compareTo(uri2)
$5 ==> 32

The expectation is that uri1.compareTo(uri2) should yield a result of 0, indicating that the URIs compare equal, but it does not. This expectation is derived from the fact that URI::compareTo is specified to satisfy the general contract of Comparable.compareTo.

Solution

Update the specification and behavior of URI::compareTo so that hexadecimal digits of escaped octets are compared without regard to case. Similar to, and consistent with, URI::equals.

With this change, the above example of uri1.compareTo(uri2) will yield a result of 0, indicating that the URIs compare equal.

Specification

src/java.base/share/classes/java/net/URI.java

   /**
    * Compares this URI to another object, which must be a URI.
    *
    * <p> When comparing corresponding components of two URIs, if one
    * component is undefined but the other is defined then the first is
    * considered to be less than the second.  Unless otherwise noted, string
    * components are ordered according to their natural, case-sensitive
    * ordering as defined by the {@link java.lang.String#compareTo(Object)
    * String.compareTo} method.  String components that are subject to
    * encoding are compared by comparing their raw forms rather than their
-   * encoded forms.
+   * encoded forms and the hexadecimal digits of escaped octets are compared
+   * without regard to case.
    *
    * <p> The ordering of URIs is defined as follows: </p>
    *
    * ...
    */
   public int compareTo(URI that) { ... }