On Tue, 13 Oct 2020 10:59:33 GMT, Hannes Wallnöfer <hann...@openjdk.org> wrote:
>> Inspired by other recent work to cleanup index generation, this change is >> primarily about cleanup for >> `AbstractIndexWriter`, `SingleIndexWriter`, `SplitIndexWriter` to reduce >> code duplication and to reduce the 3 class to >> just 1 that can handle either single index files or split index files. As >> part of this cleanup, some anomalies were >> uncovered in the naming of annotation types, enum constants and records, >> which have also been addressed. >> The changes involved moving/merging code from `SingleIndexWriter` and >> `SplitIndexWriter` in to `AbstractIndexWriter`, >> which is now no longer abstract and is thus renamed to just `IndexWriter`. >> Regrettably, the rename is not properly >> tracked. > > src/jdk.javadoc/share/classes/jdk/javadoc/internal/doclets/formats/html/IndexWriter.java > line 87: > >> 85: IndexBuilder mainIndex = configuration.mainIndex; >> 86: List<Character> firstCharacters = mainIndex.getFirstCharacters(); >> 87: firstCharacters.sort(Comparator.naturalOrder()); > > Is this sorting step necessary? Aren't the characters returned by > `IndexBuilder#getFirstCharacters()` already > guaranteed to be in natural order? Interesting ... while the contract is not enforced through the type system, the words of the spec do guarantee the ordering. ------------- PR: https://git.openjdk.java.net/jdk/pull/621
