Created attachment 177430 [details] HTML file highlighting the problem If list items from multiple lists (<ul>/<ol>) are flowed into regions, the internal list counters are not reset when moving from one list to the other. This leads to continuous numbering across all list items, regardless of their origin, list numbering style, etc. In the attached sample, the items labeled "First..", "Second..", ... are numbered 5, 6, ... where they should have been 1, 2, ...
Also, if flowing only *some* individual list items, the numbering for both the list items in the named flow and the list items in the document flow are messed up. More specifically, suppose we have two numbered lists, and extract the first <li> in a named-flow via the :first-child pseudo-class and flow them into a region. The list items left in the document will look like: 1. Item 2 2. Item 3 a. Item b b. Item c while those in the region will look like: 1. Item 1 b. Item a
Created attachment 178543 [details] HTML file highlighting the second problem If the cause for this behavior ( comment 1 ) is not the same as the initial bug, just let me know and I'll split them.
FYI: it's because the counters are calculated on the DOM and not on the RenderObjects, they have no notion that they are in the flow threads.
Created attachment 197247 [details] Patch
Comment on attachment 197247 [details] Patch Regions work correctly with lists and counters since https://bugs.webkit.org/show_bug.cgi?id=110352 . This patch only brings better test coverage for the feature.
Comment on attachment 197247 [details] Patch Clearing flags on attachment: 197247 Committed r148092: <http://trac.webkit.org/changeset/148092>
All reviewed patches have been landed. Closing bug.
Re-opened since this is blocked by bug 114519
Comment on attachment 197247 [details] Patch Re-land after http://trac.webkit.org/changeset/148863 .
Comment on attachment 197247 [details] Patch Rejecting attachment 197247 [details] from commit-queue. Failed to run "['/Volumes/Data/EWS/WebKit/Tools/Scripts/webkit-patch', '--status-host=webkit-queues.appspot.com', '--bot-id=webkit-cq-02', 'validate-changelog', '--non-interactive', 197247, '--port=mac']" exit_code: 1 cwd: /Volumes/Data/EWS/WebKit /Volumes/Data/EWS/WebKit/LayoutTests/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive). Full output: http://webkit-queues.appspot.com/results/88769
Comment on attachment 197247 [details] Patch re-r+ing
Comment on attachment 197247 [details] Patch Clearing flags on attachment: 197247 Committed r148878: <http://trac.webkit.org/changeset/148878>
This test is failing on Qt. Could you check what happened to it?a * fast/regions/counters/extract-list-items-013.html http://build.webkit.sed.hu/results/x86-64%20Linux%20Qt%20Release/r148878%20(50369)/fast/regions/counters/extract-list-items-013-diffs.html
(In reply to comment #14) > This test is failing on Qt. Could you check what happened to it?a > > * fast/regions/counters/extract-list-items-013.html > > http://build.webkit.sed.hu/results/x86-64%20Linux%20Qt%20Release/r148878%20(50369)/fast/regions/counters/extract-list-items-013-diffs.html Baaaah :(. it's that font smoothing bug again. I'd like to do a smarter fix for this so is it possible for you to disable the test for now? I'll make sure to re-enable once I land a proper fix for all of the ref-tests we have.