User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.4; en-GB; rv:1.9.2.13) Gecko/20101203 Firefox/3.6.13 Build Identifier: Firefox currently allows line breaking before a punctuation mark when it is preceded by a breakable space, such as a thin space U+2009 (as used in French). This is incorrect. The line breaking algorithm should be improved, for instance as described by: http://www.unicode.org/reports/tr14/ (Unicode Standard Annex #14 - Unicode Line Breaking Algorithm). In particular, see LB13: "Do not break before ‘]’ or ‘!’ or ‘;’ or ‘/’, even after spaces." Note that the list of cited characters is not exhaustive. Reproducible: Always Steps to Reproduce: 1. Open a page with a word followed by a thin space then by a punctuation mark such as "!" or "?" (see testcase). 2. Adjust the window width to see the various line breaking positions. Actual Results: For some width, Firefox breaks the line before the punctuation mark. Expected Results: Firefox should break the line before the word preceding it. WebKit-based browsers have no such problem (tested only with a thin space). The normal space should be handled as well, because some authors use it instead of a thin space (even with such a slightly incorrect typography, authors often don't use the typical no-break space, probably because a normal space is easier to type on a keyboard).

See also Bug 56652 and Bug 206152 with its Dependencies.

Last I checked we purposefully allowed breaking before '/' because it's common for sites to depend on that.

How about the importance of this bug for Western language users? Is that a major format?

Hmm, looks like the spaces are marked as breakable in nsLineBreaker::AppendText(). However, we need to access to nsILineBreaker's implementation when we check whether the next character is open punctuation or not...

s/open punctuation/close punctuation

http://mxr.mozilla.org/mozilla-central/source/intl/lwbrk/src/nsJISx4501LineBreaker.cpp#867 Now, GetJISx4051Breaks() always returns FALSE for first character. It might be usable. If the first character is not breakable even with a breakable character, we can set TRUE. Then, nsLineBreaker can know the first character's class of "word" and it can overwrite the previous space's state. But this is just my guess.