28237 2015-03-19 08:50:11 +0000 [WebVTT] Should :past and :future pseudo-classes be each others negations? 2015-04-19 03:15:46 +0000 1 1 1 Unclassified TextTracks CG WebVTT unspecified PC All NEW http://dev.w3.org/html5/webvtt/ P2 normal --- 1 philipj dave.null eric.carlson fs philipj silviapfeiffer1 zcorpan public-texttracks oldest_to_newest 118674 0 philipj 2015-03-19 08:50:11 +0000 As implemented in Blink and WebKit, :past is equivalent to :not(:future) and vice versa. This is not per spec, where :past and :future will only match if there's a timestamp node somewhere in the cue. What do other implementations do? What is most useful for developers? (Note that Blink's implementation makes it possible to only traverse the nodes once, where the spec's implementation would make it moderately more complex.) 118675 1 philipj 2015-03-19 08:55:20 +0000 Fixed title to be less wrong. :past and :future are mutually exclusive per spec as well, but it's possible for neither to match. 118678 2 zcorpan 2015-03-19 11:55:46 +0000 (In reply to Philip Jägenstedt from comment #0) > What is most useful for developers? See https://www.w3.org/Bugs/Public/show_bug.cgi?id=16875 118679 3 philipj 2015-03-19 12:12:43 +0000 (In reply to Simon Pieters from comment #2) > (In reply to Philip Jägenstedt from comment #0) > > What is most useful for developers? > > See https://www.w3.org/Bugs/Public/show_bug.cgi?id=16875 Any comment in particular? I agree that it's a problem, more serious than this bug, but it they seem like orthogonal problems to me. 118682 4 zcorpan 2015-03-19 13:10:20 +0000 I was thinking that fixing that bug might make this bug moot, but it depends on what the fix is. 118702 5 philipj 2015-03-20 05:38:12 +0000 Eric, WebKit is also affected by this. Fredrik and I think doing what the spec currently says seems fine, does that sound OK to you?