On Tue, 11 Jun 2024 23:49:15 GMT, Jonathan Gibbons <j...@openjdk.org> wrote:
>> As @jonathan-gibbons likes to point out, javadoc is not an HTML validation >> tool. So I think, it's okay to leave the code simple. Maybe this would be >> even simpler? >> >> data-[a-z][-a-z0-9]* > > I admit to being "lazy" when I wrote `startsWith("on")`. In general, I don't > think folk should be writing many if any event handlers in doc comments, and > it was easier to give forward compatibility with a weak test for `"on"`. > Likewise, there should not be many if any uses of `data-*` attributes, and it > is easy enough to give a simple regex. It's more important to not reject > valid uses of event attributes and custom data attributes than it is to > detect all invalid cases. The right place to detect all invalid cases is > downstream in recognized validation tools, like `tidy` which catch all > occurrences of bad stuff, wherever they originate (doc comment, command-line > options, the `javadoc` tool, post-processors, etc). DocLint is primarily an > "early warning system" for common errors. I agree with you, Jon. So, we can simply use `startsWith("data-")`, which would be good enough and most permissive (i.e. no false negatives). ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/19652#discussion_r1636126138