Bug 27074 – Text should wrap after hyphens, not before

• Component: Layout and Rendering
• Status: RESOLVED WORKSFORME

Calibre (http://calibre.kovidgoyal.net/) uses Webkit, I'm not sure what version, and it wraps hyphens to the next line, which looks weird, and not book-like. The dev says it's a WebKit issue, so I'm reporting it here.

[Mark Rowe (bdash)]

Perhaps you'd like to include a list of steps necessary to reproduce the problem?

1. Download and install Calibre (http://calibre.kovidgoyal.net/download).
2. Open an ebook (http://www.epubbooks.com/book/60/count-of-monte-cristo) with Calibre's viewer.
3. Resize the window until it has to break a hyphenated word (of which this book has several right at the top).

Of the few ebooks I've looked at, only this one puts the hyphen on the new line. I see nothing suspicious in the HTML, so I don't know why.

[Alexey Proskuryakov]

> I see nothing suspicious in the HTML, so I don't know why.

Is this reproducible with WebKit-based Web browsers? Could you make a reduction, or upload the HTML source here?

In looking at the source, I see that the hyphens are actually en-dashes misused. I'm no typographer, but the only place I've know an en-dash to occur without spaces around it (other than this misuse as a hyphen) is in a range, such as 2–5. Should breaking occur at all at an en dash?

[Xianzhu Wang]

Created attachment 71384 [details]
test case

I couldn't reproduce the bug in Chrome 7 and 8 and Safari 5, but I could reproduce it in Qt version of WebKit.

Qt version of WebKit uses QTextBoundaryFinder to find line breaking opportunities, so this bug should be of QT, not of WebKit itself.

[Pat]

Using QtWebKit trunk (wk45), this is not reproducible.
The test content is using entities ߝ for the endash and the breaks occur *after* the dash. Tried the test content saved as UTF-8 with and without BOM, also UTF-16 and the content displayed and broke after the endash. Used QtTestBrowser to verify on Symbian and Linux. Both work as expected.

Test content, as provided, used:
<!DOCTYPEhtml>
abcdefg&#x2013;abcdefg&#x2013;abcdefg&#x2013;abcdefg&#x2013;abcdefg&#x2013;abcdefg&#x2013; ....


Note, that test content has no meta tag:
<meta http-equiv="content-type" content="text/html; charset=UTF-16"> 
Tested with meta tag added and works as expected.

The behavior is the same for Firefox, Safari, and QtTestBrowser.