Tuesday, 2015-06-23

*** sballe has quit IRC04:02
*** kes_ has joined #openstack-searchlight12:31
*** kes_ has quit IRC13:17
*** openstack has quit IRC13:17
*** openstack has joined #openstack-searchlight13:20
*** sigmavirus24_awa is now known as sigmavirus2414:07
*** kexiaodong has joined #openstack-searchlight14:33
kexiaodonganyone here? I hava a question about searchlight. Why we only choose Elasticsearch? Why not use driver to config search layer? We can use solr, or mongodb, or something else.14:35
sjmc7that's a possibility for the future, but we rely pretty heavily on elasticsearch's features. we might one day add the ability to use other backends but we won't be writing a query abstraction language14:38
sigmavirus24we're also very very young14:43
sigmavirus24so we're focusing on functionality over configurability14:43
sigmavirus24Searchlight needs to work before we want to add search backend drivers14:44
*** sballe has joined #openstack-searchlight14:44
kexiaodongCan we make a query layer? I have 2 more reasons: (1) the query language(dsl) abort elasticsearch is very hard to use, (2) I think we can cover most case by one or two parameter in full context search, so we can make search easily.14:45
sigmavirus24kexiaodong: we're not going to work on that right now. Our first priority is to get a version of Searchlight that works as advertised and can be deployed for Liberty14:46
sigmavirus24And right now, not making our own query language means that goal is far more achieveable14:47
sigmavirus24We will not be bikeshedding over what the query language will look like. We will not be bike shedding over how closely to tie the query language to Elasticsearch or limit it by other backend's abilities14:47
kexiaodongok14:47
sigmavirus24That can all come in the future, once Searchlight has had a stable release with real adoption14:47
sigmavirus24Without users, we can't know what people will really need from us14:48
sigmavirus24Horizon is already one very enthusiastic consumer, but we don't want to just tailor to Horizon and we can't do that until we have a release14:48
sjmc7i also don't think that we'll be able to come up with a query language that's easier to use that doesn't lose a lot of flexibility14:48
sjmc7what we may do is write a client wrapper to simplify common queries14:48
sigmavirus24sjmc7: agreed14:48
nikhil_ksjmc7: no l1 content then?15:57
sjmc7?15:58
nikhil_kI don't think the rel-mgrs are aware or care about this project yet for releases. so just trying to communicate  to them and amke them aware just in case15:58
nikhil_ksjmc7: ^15:58
nikhil_kand I was curious if we should go with something out in l2?15:58
sjmc7ah, i see. yeah, i'm ok with no l1, but i think we should aim for l2 if we want a liberty release15:59
nikhil_kThe releases are being cut today through thursday15:59
nikhil_ksjmc7: cool. sounds like plan.15:59
sjmc7ok. i don't know what's involved with it, but i suspect 0-days notice is not the way to endear ourselves to them :)15:59
sjmc7can we get them to do a test build sometime before l2?16:00
nikhil_kseems like we may get to tag our own release16:05
nikhil_ksjmc7: so, no worries for now :) But we may just get a bit complacent16:05
nikhil_kI will try to push myself in seeing one through in July < o >16:06
sjmc7we'll all push :)16:10
*** dontalton has joined #openstack-searchlight17:01
*** openstackgerrit has quit IRC19:21
*** openstackgerrit has joined #openstack-searchlight19:22
*** dontalton has quit IRC20:32
*** dontalton has joined #openstack-searchlight20:33
kexiaodongSorry out, I need see through the full source to join more.21:01
*** dontalton2 has joined #openstack-searchlight22:20
*** sigmavirus24 is now known as sigmavirus24_awa22:26
*** dontalton2 has quit IRC23:16
*** dontalton has quit IRC23:16
*** kexiaodong1 has joined #openstack-searchlight23:50
*** kexiaodong1 has quit IRC23:52

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!