[18:55:12] <jeremyb-phone>	 is there a youtube link for elasticsearch?
[19:03:35] <bd808>	 jeremyb-phone: Not yet. I'll post as soon as I find one.
[19:03:50] <bd808>	 jeremyb-phone: There is just https://plus.google.com/u/0/events/cokipb2senmmvkvdjif7aq55kac for now
[19:05:30] <bd808>	 jeremyb-phone: https://www.youtube.com/watch?v=FubXExbAvOA
[19:12:57] <jeremyb>	 (thanks, also following in #-dev)
[19:48:50] <cscott>	 pasting some comments on the talk from #mediawiki-parsoid:
[19:48:52] <cscott>	 (12:46:04 PM) subbu: cscott, maybe you can follow up on nik's response ... but curious how will a html document store change the search issues? or, maybe this is an offline discussion.
[19:48:52] <cscott>	 (12:46:22 PM) subbu: or actually, maybe it won't.
[19:48:52] <cscott>	 (12:46:25 PM) subbu: never mind.
[19:48:52] <cscott>	 (12:46:38 PM) subbu: in the sense it is still just a api call as far as cirrus is concerned.
[19:48:55] <cscott>	 (12:46:46 PM) cscott-free: yeah, i wanted to figure out whether or not their troubles getting good sentence data was due to using non-semantic HTML output from the PHP parser
[19:48:56] <cscott>	 (12:47:05 PM) cscott-free: for example, our <figure> markup should be easier to identify and separate out
[19:48:57] <cscott>	 (12:47:15 PM) subbu: i think so.
[19:48:58] <cscott>	 (12:47:39 PM) cscott-free: probably table markup would be comparable, so no difference there
[19:48:59] <cscott>	 (12:47:49 PM) cscott-free: but we could allow some more precise info on templates
[19:49:00] <cscott>	 (12:48:03 PM) cscott-free: maybe lowering the relevance of hits which occur inside templates, for example
[19:49:02] <cscott>	 (12:48:13 PM) cscott-free: that's lost in PHP HTML output