88749 2012-06-10 21:30:21 -0700 HTMLSelectElement::selectedOptions returns wrong collection 2012-06-27 18:36:15 -0700 1 1 1 Unclassified WebKit DOM 528+ (Nightly build) All All RESOLVED FIXED http://jsfiddle.net/f39cC/5/ InRadar, WebExposed P2 Normal --- 1 tkent webkit-unassigned ap benjamin jonlee menard michelangelo mitz simon.fraser tonikitoo webkit-bug-importer oldest_to_newest 645717 0 tkent 2012-06-10 21:30:21 -0700 http://code.google.com/p/chromium/issues/detail?id=131890 1. Open http://jsfiddle.net/f39cC/5/ 2. Select One, then add Two to the selection, then Three 3. Select Two only What is the expected result? The following results are respectively expected: "One", "One,Two", "One,Two,Three", "Two". What happens instead? We get "One", "One", "One,Two", "Two,,". 649930 1 tkent 2012-06-14 23:47:20 -0700 I'll roll out http://trac.webkit.org/changeset/110340 because this feature is incomplete at all. For example, If we change select-selectedOptions.html so that debug((++i) + ") Select an option should update the selected options collection."); mySelect.options[0].selected = true; shouldBe("mySelect.selectedOptions.length", "1"); // reorder these two lines shouldBe("mySelect.options.length", "2"); // ditto. shouldBe("mySelect.selectedOptions[0].text", "'one'"); the test fails. HTMLCollection invalidates the content if domTreeVersion is changed. But selected state change doesn't update domTreeVersion. The test works accidentally because shouldBe() updates domTreeVersion. 649951 2 tkent 2012-06-15 00:13:51 -0700 Committed r120413: <http://trac.webkit.org/changeset/120413> 650199 3 webkit-bug-importer 2012-06-15 08:13:20 -0700 <rdar://problem/11676534> 657981 4 simon.fraser 2012-06-26 14:15:52 -0700 This rollout broke Garden-o-matic. 657983 5 menard 2012-06-26 14:18:38 -0700 The rollout is overkill. I can just fix the bug. Now I will need to land again the feature + fix it :(. 658194 6 menard 2012-06-26 18:09:39 -0700 (In reply to comment #5) > The rollout is overkill. I can just fix the bug. Now I will need to land again the feature + fix it :(. For the record : http://jsfiddle.net/f39cC/20/ works fine. What seems to get crazy is when using : output.innerHTML = ([]).map.call(this.selectedOptions, function(so) {return so.value;}).join(); console.log(([]).map.call(this.selectedOptions, function(so) {return so.value;}).join()); together. Using one of them at a time is fine. "because this feature is incomplete at all." Well we just spotted a bug, but as the jsfiddle shows the basic use case work. We do have a problem I agree with that and it's related to the selected state not invalidating properly the selectedOptions collection. Also I couldn't reproduce with the suggestion of comment #1 unless I missed something. I do have a patch here to fix the original issue, I'll cleanup and upload tomorrow. 658208 7 tkent 2012-06-26 18:27:47 -0700 I'm sorry for rolling out. I was too aggressive. I wanted to resolve this issue before Chromium M21 branch, and the bug had no response for four days. 658227 8 menard 2012-06-26 18:39:29 -0700 (In reply to comment #7) > I'm sorry for rolling out. I was too aggressive. I wanted to resolve this issue before Chromium M21 branch, and the bug had no response for four days. I'm on IRC, don't hesitate to ping me. For some reason you CC me but it was in my "to read" bugzilla folder, sorry to have missed the email. 659071 9 menard 2012-06-27 18:36:15 -0700 (In reply to comment #8) > (In reply to comment #7) > > I'm sorry for rolling out. I was too aggressive. I wanted to resolve this issue before Chromium M21 branch, and the bug had no response for four days. > > I'm on IRC, don't hesitate to ping me. For some reason you CC me but it was in my "to read" bugzilla folder, sorry to have missed the email. I fixed the bug.