Putting an empty span in the middle of a word should not make the word breakable at the span position. See this page for an example: http://www.plexode.com/cgi-bin/eval3.py#ht=%3Cdiv%20style%3D%22background-color%3Ared%3Bwidth%3A1px%22%3Ewo%3Cspan%3E%3C%2Fspan%3Erd%3C%2Fdiv%3E&ohh=1&ohj=1&jt=&ojh=1&ojj=1&ms=100&oth=0&otj=0&cex=1 I'm not sure why this is correct, but roc claims it is and this is what IE7 and FF3 do.
The CSS spec does not define linebreaking in enough detail to say whether this bug is valid. However, I think as a matter of principle inserting unstyled <span>s in and around inline content should not change rendering.
Created attachment 205514 [details] Patch
Created attachment 205822 [details] Patch
Comment on attachment 205822 [details] Patch r=me
Comment on attachment 205822 [details] Patch Attachment 205822 [details] did not pass win-ews (win): Output: http://webkit-queues.appspot.com/results/1018193 New failing tests: fast/js/global-constructors-attributes-shared-worker.html fast/forms/select/popup-closes-on-blur.html fast/media/mq-transform-03.html editing/selection/user-select-all-image-with-single-click.html fast/forms/search-event-delay.html fast/js/global-constructors-attributes-dedicated-worker.html editing/selection/user-select-all-with-single-click.html fast/media/mq-transform-02.html platform/win/accessibility/multiple-select-element-role.html
Created attachment 205930 [details] Archive of layout-test-results from APPLE-EWS-3 for win-future The attached test failures were seen while running run-webkit-tests on the win-ews. Bot: APPLE-EWS-3 Port: win-future Platform: CYGWIN_NT-6.1-WOW64-1.7.20-0.266-5-3-i686-32bit
Committed r152313: <http://trac.webkit.org/changeset/152313>
This caused a regression here: bug 118435