> Instances of TypeMirror that are equal (TypeMirror::equals), aren't > necessarily the same (Types::isSameType). If care is not taken when putting > instances of TypeMirror into a set, that set might end up containing the same > instances. > > If Utils.getAllInterfaces is called on a type that extends or implements a > particular interface multiple times (on different levels of that type's > hierarchy), the returned set might contain multiple representations of that > interface. For example, I've seen a case where getAllInterfaces that was > passed a TypeElement corresponding to java.util.ArrayList returned a set > containing 3 instances of TypeMirror corresponding to > `java.util.Collection<E>`. > > A bit of archaeology. The old standard doclet (removed in JDK-8177511, commit > 33ab1995) collected instances of com.sun.javadoc.Type in a TreeMap, keying > them by instances of com.sun.tools.javadoc.main.ClassDocImpl (removed in > JDK-8215584, commit 151e628) which implemented Comparable. > ClassDocImpl.compareTo worked by comparing instances of CollationKey derived > from FQNs of types represented by respective instances of ClassDocImpl. > > I'm not sure why `TreeMap<ClassDocImpl, Type>` was changed to > `LinkedHashSet<TypeMirror>`.
Pavel Rappo has updated the pull request incrementally with one additional commit since the last revision: Add comments Explain the rationale and the mechanics of the changes so far. ------------- Changes: - all: https://git.openjdk.java.net/jdk/pull/7659/files - new: https://git.openjdk.java.net/jdk/pull/7659/files/a7d3ca0b..00259e26 Webrevs: - full: https://webrevs.openjdk.java.net/?repo=jdk&pr=7659&range=02 - incr: https://webrevs.openjdk.java.net/?repo=jdk&pr=7659&range=01-02 Stats: 87 lines in 1 file changed: 83 ins; 0 del; 4 mod Patch: https://git.openjdk.java.net/jdk/pull/7659.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/7659/head:pull/7659 PR: https://git.openjdk.java.net/jdk/pull/7659
