Looks good to me. Hannes
> Am 14.12.2019 um 03:17 schrieb Jonathan Gibbons <[email protected]>: > > Please review a moderately simple cleanup to the implementation(s) of > Utils.getBlockTags. > > The existing code is unnecessarily string-oriented, and can be improved by > better leveraging DocTree.Kind, especially by updating each subtype of > BaseTaglet to know its associated DocTree.Kind. > > The core of the fix is Utils, with additional support in BaseTaglet and > SimpleTaglet. The other changes are derivative changes using the new API. > > There are more changes possible in this (general) area. For example, there > are similar methods such as Utils.hasBlockTag, and methods like > CommentHelper.getTagName. At a minimum, it may be reasonable to co-locate all > these methods in a new "Tags" utility class, but it is also worth > investigating what additional simplifications can be made. But for now, this > is a good checkpoint. > > The old code accidentally covered up a pre-existing bug, which was exposed in > the replacement code. The old code did not return @uses and @provides from > getBlockTags, and so they did not not to be skipped as part of the main > comment in ModuleWriterImpl. Now they are returned by getBlockTags, and so > need to be skipped in TagletWriter. > > This is all cleanup with no changes in the generated output. There are no new > tests and no changes needed to any existing tests. A full comparison against > a reference JDK was done with the standard JDK docs (make docs) and with all > the output from all the jtreg javadoc tests. > > -- Jon > > JBS: https://bugs.openjdk.java.net/browse/JDK-8235947 > Webrev: http://cr.openjdk.java.net/~jjg/8235947/webrev.00/ >
