On Sat, May 17, 2014 at 7:08 PM, Russavia <russavia.wikipedia(a)gmail.com>wrote;wrote:
I understand the new search is currently being worked
on and refined,
and will obviously be rolled out across all projects in a timeframe I
am not too sure of. Can we get that timeframe/update on where this
project is at, and when those working on it expect it to be "stable".
Once new search is working, the first enhancement to the search should
be a clustering feature. Wouldn't such a feature pretty much solve
the problem that we currently have with search, and which won't be
solved by the "out-of-the-box" search that is being worked on now.
Just for a quick status update on the search project for those who might
not be following it.
We're currently live on all non-Wikipedias, non-Commons, non-Meta and
non-Incubator as the primary search engine and have been for a little while
now. The Wikipedias and the 3 misc. projects I mentioned above all have
the new search available as a Beta Feature. We've had many thousands of
users trying things out and the feedback has been very positive thus far.
(shameless plug to please give feedback on [[mw:Talk:Search]] or Bugzilla.
Nothing helps us find bugs and get them fixed faster than user feedback)
We could probably swap the remaining non-English Wikipedias, Meta
and Incubator into having the new search engine as their primary without
worrying about performance. Commons too, possibly, but we'd have to
keep an eye on things. enwiki we know we can't quite handle yet but we're
working on it.
Vague timelines suck I know, but trying to get the performance we need
out of Elasticsearch is a multifaceted problem and we've been trying to
roll this out with the minimal amount of disruption to everyone as possible.