On Fri, 22 Nov 2024 14:36:39 GMT, Nizar Benalla <nbena...@openjdk.org> wrote:
>> Please review this patch to improve the javadoc preview header. The parent >> bug include more details. >> >> This is meant to prevent a bug where, if a non preview class extend an >> `@PreviewFeature` interface, the Javadoc displays `Programs can only use >> <<Type>> when preview features are enabled` when the API can be used without >> enable preview features as long as we don't refer to any methods in the >> supertype >> >> This is specifically the case in #17543. Here is the javadoc output >> [AsymmetricKey](https://cr.openjdk.org/~nbenalla/GeneratedDocs/8343239-2/docs/api/java.base/java/security/AsymmetricKey.html) >> after merging this branch with the PR `298420: PEM API: Implementation >> (Preview) #17543`. The only files affected by this change are the class >> `DEREncodable` and it's Subinterfaces. >> >> One other bug I noticed is that, on the `DEREncodable` page, it's direct >> subinterfaces had the superscript `PREVIEW` link. I've added a condition to >> remove it. > > Nizar Benalla has updated the pull request with a new target base due to a > merge or a rebase. The incremental webrev excludes the unrelated changes > brought in by the merge/rebase. The pull request contains five additional > commits since the last revision: > > - dropping earlier idea/message as non-preview class should not have a > preview message. As you do not need --enable-preview to use them. > > Removed some code and modified test accordingly > - Merge remote-tracking branch 'upstream/master' into javadoc-preview-message > - remove unused import > - improve preview label message > - fix preview bug src/jdk.javadoc/share/classes/jdk/javadoc/internal/doclets/formats/html/HtmlDocletWriter.java line 2519: > 2517: target.add(previewDiv); > 2518: } else if ((forWhat.getKind().isClass() || > forWhat.getKind().isInterface()) > 2519: && !utils.nonPreviewExtendsPreview(forWhat)) { It appears you are excluding a type from getting a preview notice by checking a condition that is elsewhere used to decide it needs a preview notice. IMO the better solution would be to not give it a preview notice in the first place by not checking implemented interfaces for classes in `Utils.declaredUsingPreviewAPIs(Element)`. (An interface extending a preview interface is a different again, although I don't think that should ever occur in the JDK.) That leaves open the case where a preview interface introduces a default method (which is being fixed for javac in [JDK-8343540](https://bugs.openjdk.org/browse/JDK-8343540)), but I would consider that a separate issue. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/22126#discussion_r1857231522
