This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
In §6.2: "Regardless of the value of the 'white-space' property, any line breaks inserted by the user agent for the purposes of line wrapping must be placed so as to minimize Δ across each run of consecutive lines between preserved newlines in the source. Δ for a set of lines is defined as the sum over each line of the absolute of the difference between the line's length and the mean line length of the set." This specifies a single line wrapping strategy, but not one that is appropriate for all usages even in horizontal scripts. Specifically, it is common authorial practice to use grammatical analysis to achieve a better reading experience even at the expense of balanced length lines. Usually this is done by a human, I believe, but implementations should be permitted to attempt it too. In §1.1: "In general, therefore, authors are encouraged to write cues all on one line except when a line break is definitely necessary, and to not manually line-wrap for aesthetic reasons alone." I would resolve this by 1) permitting other line wrapping strategies explicitly in §6.2, i.e. making the above clause optional and possibly parameterised, and 2) by amending the encouragement in §1.1. It is unclear what is meant by the words "aesthetic reasons alone" here. A possible alternate wording is: "In general authors are encouraged either to insert line breaks explicitly or to write all cues on a single line, and authors are reminded that even if line breaks are inserted explicitly, additional line breaks could be inserted to attempt to render all of the text, depending on space available, amount of text and font size."
Related to bug 28267 and bug 17218 and even bug 19458.
I disagree that we should allow for any differences between implementations at all, but it may very well be that the line balancing described is no good and that we should change the required default to something else. Since nobody has implemented it, perhaps we should just specify what everyone has actually implemented.