On Thu, 25 Jan 2024 23:44:06 GMT, Andy Goryachev <ango...@openjdk.org> wrote:
>>> Regarding other whitespace characters. Using \u2003 em space, pasting the >>> following text >> >> I think we can use `Character.isWhitespace`. For the most part it aligns >> with "is this a breaking space". It's not perfect, but we can do more when >> there are complaints. Here's my test: >> >> >> char isWhiteSpace (breaking space)? >> \u0009 true >> \u0020 true >> \u00a0 false >> \u1680 true >> \u180e false <- unexpected >> \u2001 true >> \u2002 true >> \u2003 true >> \u2004 true >> \u2005 true >> \u2006 true >> \u2007 false >> \u2008 true >> \u2009 true >> \u200a true >> \u200b false <- unexpected >> \u200c false <- unexpected >> \u200d false <- unexpected >> \u202f false >> \u205f true >> \u3000 true > > thank you @hjohn ! this PR has been out for a while, could you sync it up > with the master please? @andy-goryachev-oracle thanks for testing this, what you discovered was a bug in the `computeTrailingWhiteSpace` code that went undetected because of the simplicity of the strings I tested with. The `positions` array containing all the character widths is shared amongst all `TextRun`s (it does mention that) but I accessed it as if it was 0 based per `TextRun`. This means it sometimes included the width of characters that weren't spaces at all... ------------- PR Comment: https://git.openjdk.org/jfx/pull/1236#issuecomment-1911261917
