If an element has the -webkit-flow-from property set, then the pseudo-elements ::before and ::after are not displayed. Trying to get the computed style for them returns a CSSStyleDeclaration object, but its properties do not have the values specified in CSS. Namely, the 'content' property has the value equal to the name of the flow that flows in the region, instead of what's specified via the 'content' declaration in CSS.
Created attachment 129906 [details] Ref test highlighting the problem
Still repros on today's nightly (r151543)
Created attachment 204749 [details] Proposed patch patch
Comment on attachment 204749 [details] Proposed patch r=me
Comment on attachment 204749 [details] Proposed patch Clearing flags on attachment: 204749 Committed r151647: <http://trac.webkit.org/changeset/151647>
All reviewed patches have been landed. Closing bug.
Comment on attachment 204749 [details] Proposed patch View in context: https://bugs.webkit.org/attachment.cgi?id=204749&action=review > LayoutTests/platform/mac/fast/regions/region-dynamic-after-before-expected.txt:1 > +layer at (0,0) size 800x600 These tests should be ref tests (the expected result should be an html file that renders the same as the test but without using regions). Please file another bug to change the expected results of these new tests.
Hi Igor, Why did you need { position: absolute, top: 0, left: 0 } for before and { position: absolute, bottom: 0, right: 0 } for after in your tests? Looking at your patch, i doubt that it follows the processing model for pseudo-elements, as described http://dev.w3.org/csswg/css-regions/#pseudo_elements. Have you tested the regions pseudo-elements with different display values?
Hi Mihai, display: run-in is a totally different beast, specially because run-in + generated content open a can of worms. I will open a different bug for css region processing model. About the the current bug tests case, i will change it. (In reply to comment #8) > Hi Igor, > > Why did you need { position: absolute, top: 0, left: 0 } for before and { position: absolute, bottom: 0, right: 0 } for after in your tests? Looking at your patch, i doubt that it follows the processing model for pseudo-elements, as described http://dev.w3.org/csswg/css-regions/#pseudo_elements. Have you tested the regions pseudo-elements with different display values?