>>>>> On Thu, 12 Sep 2024 13:53:26 -0400, John Ciolfi via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org> said:

    John> 1. Create foo.xml containing:

    John> <foo>
    John>     <!--
    John>         foo

    John> bar
    -->
    John> </foo>

    John> Notice that we have a blank line in the xml comment.

    John> 2. Put the point on 'bar' and press TAB.

    John> Notice that the 'bar' line is not indented correctly.

    John> Solution
    John> --------

    John> The problem is that in nxml-compute-indent-in-delimited-token, there
    John> is a (forward-line -1) which assumes that the prior line is indented. In
    John> this case the prior line is a blank line and thus doesn't have
    John> indentation. After the (forward-line -1), we need to keep going back
    John> until we see a non-blank line:

    John>           (forward-line -1)
    John> +           (while (looking-at "^[:blank:]*$")
    John> +          (forward-line -1))
 
    John> Attached is a patch for nxml-mode.el that fixes the issue.

Your patch does the right thing, but for the wrong reasons. Regexp
character classes only have meaning with a bracket expression, so what
you’re matching here is actually any lines containing zero or more of
':', 'b', 'l', 'a', 'n', or 'k'. IOW, you should use:

     (looking-at "^[[:blank:]]*$")

Robert
--