The 'word-break' property supports 5 values, as declared in CSS3 Working Draft 6. http://www.w3.org/TR/css3-text/#word-break The 'word-break' property is crucial in order for wrapping Asian text properly. However, the latest Webkit engine fails to recognize some of 'word-break' property's values, for instance the 'keep-all' value. It is apparent as it can be viewed on the inspector, that Webkit fixes unsupported values into 'normal'.
Do you have an example of a site that breaks, and/or a simple test case?
Created attachment 64393 [details] Test Page A test page on word-break: keep-all; XHTML validated and ready to go. The first paragraph contains Korean letters(Hangul), one of the Asian text that word-break was targeted for. If a Korean font is installed, it'll be noticed that currently Korean text breaks anywhere between letters. There's also a identical copy of this file on http://juneoh.net/word-break.html for viewing.
Confirmed with r65392 on Mac OS X.
(In reply to comment #0) > The 'word-break' property supports 5 values, as declared in CSS3 Working Draft 6. > http://www.w3.org/TR/css3-text/#word-break > The 'word-break' property is crucial in order for wrapping Asian text properly. > > However, the latest Webkit engine fails to recognize some of 'word-break' property's values, for instance the 'keep-all' value. > It is apparent as it can be viewed on the inspector, that Webkit fixes unsupported values into 'normal'. Do you mean 'line-break' not 'word-break'? According to the CSS3 spec., the 'work-break' property has just 3 values such as 'normal', 'break-all', and 'hyphenate'.
Dear Joone Hur, this issue was filed when the previous Working Draft of CSS3(http://www.w3.org/TR/2007/WD-css3-text-20070306/) was active. I see now that the latest version of the draft has removed the value keep-all from the word-break property and passed it on to line-break. I do not know if the problem recurs with the same value on line-break property, so I would appreciate if someone could reconfirm the issue in regard to this change.
Created attachment 106097 [details] word-break test case In the current working draft 'keep-all' is a property of 'word-break'. This is also relevant to non-Asian text. Attached is a simpler test case, where webkit splits between the minus sign and the number with both the normal and keep settings, whereas Firefox never introduces a break there. This relates to bug 20677.
I added a patch for this on bug 123782
Safari, Chrome, and Firefox show the same rendering behavior for this test case. I do not believe any further compatibility issue remains.