-
legoktm
oh boy, renaming users
-
that's always a blast
-
central4 has quit
-
d4rkie joined the channel
-
D4RK-PH0ENiX has quit
-
D4RK-PH0ENiX joined the channel
-
d4rkie has quit
-
D4RK-PH0ENiX has quit
-
D4RK-PH0ENiX joined the channel
-
Leo__Verto has quit
-
Slurpee has quit
-
sentriz has quit
-
sentriz joined the channel
-
UmkaDK joined the channel
-
samj1912
Morning people o/
-
And evening Major_Lurker o/
-
Major_Lurker
good evening samj1912
-
ha you beat me
-
obtext joined the channel
-
samj1912[m] joined the channel
-
UmkaDK has quit
-
UmkaDK joined the channel
-
UmkaDK_ joined the channel
-
UmkaDK has quit
-
ruaok
\ø moooin!
-
CatQuest
yello
-
iliekcomputers
moin!
-
d4rkie joined the channel
-
D4RK-PH0ENiX has quit
-
samj1912
ruaok, can you take a look the reminding sir PRs?
-
*remaining
-
drsaunder has quit
-
ruaok
-
samj1912
Yup
-
I tested it out, seemed to work well for things like Sony and Universal
-
Freso
samj1912: Does refcount use relationships too?
-
samj1912
For some it does
-
Freso
A lot of labels are specifically not for Release Label use, but are still used a lot (e.g., n©b, GEMA, etc.). Not sure if it'll have any effect, but if I write "ncb", I'd expect n©b to pop up first. Is it still live on test for testing?
-
samj1912
yup
-
Freso
«yes, feel free to wreak havoc» you mean? :D
-
samj1912
-
just use the advanced query method since only that uses the right boosts currently
-
Freso
👍
-
samj1912
-
Freso
-
samj1912
we are now returning just a count instead of returning all the ids and then counting them in python
-
Freso: I dont think the ref count part is live yet for labels
-
Freso
Ah, okay.
-
ruaok
samj1912: got it -- makes sense.
-
github joined the channel
-
github
[sir] samj1912 closed pull request #79: Improve query efficiency when calculating ref_counts (master...count)
https://git.io/vppNk
-
github has left the channel
-
samj1912
hmm, this is new, GH now removes PR approvals if you push/rebase
-
which kinda makes sense
-
github joined the channel
-
github
[sir] samj1912 closed pull request #78: Add release counts for labels (master...label)
https://git.io/vpxhl
-
github has left the channel
-
iliekcomputers
ruaok: re the GDPR thingy what do you think we should do, add logic for not creating an account if I disagree is ticked?
-
drsaunder joined the channel
-
(I'm not sure why anyone without an account would login to go to the deletion page)
-
samj1912
Freso: now the refcount for labels should be live
-
Tykling has quit
-
reosarevok: ping
-
Tykling joined the channel
-
hmm, seems liek hes not here
-
Freso: you wanna help :P?
-
reosarevok
samj1912: I'm out for a while longer, I'll be home in an hour or so I guess
-
If you have some short questions I can take a look, but no continuous checking rn :)
-
Freso
-
samj1912
reosarevok: wanted to tweak other entities
-
lets do that when you are home
-
D4RK-PH0ENiX joined the channel
-
zas
alastairp: are you ready to move acoustibrainz stuff from spike to another server ?
-
d4rkie has quit
-
D4RK-PH0ENiX has quit
-
drsaunder has quit
-
Leo__Verto joined the channel
-
D4RK-PH0ENiX joined the channel
-
D4RK-PH0ENiX has quit
-
Slurpee joined the channel
-
D4RK-PH0ENiX joined the channel
-
reosarevok
samj1912: around
-
samj1912
reosarevok: hey
-
wanna do places?
-
ruaok
iliekcomputers: hmmm. I've been thinking about this and I'm really not sure how to proceed.
-
lets make sure we see the same problem first.
-
if I read the PR correctly, if they agree, they log in and all good.
-
if they disagree, then we log them in and take them to the delete my account page.
-
reosarevok
samj1912: sure :)
-
ruaok
from there, the can just go about their business and never have agreed to the GDPR flag.
-
samj1912
-
ruaok
I think it is a bad idea to just log the user in if they do not agree to the GDPR flag.
-
samj1912
this is the current state
-
it currently has no boosts whatsoever
-
address alias area comment ngram pid place type
-
I guess we might want to boost the place and alias
-
to 2 and 1.5?
-
anything else?
-
reosarevok
Area is important
-
samj1912
okay, so 1.5 to that?
-
or 1.2?
-
reosarevok
Maybe try that 1.5 too, yes
-
samj1912
can you do a test query
-
reosarevok
LEt's see if that makes a mess, if it does, 1.2
-
samj1912
then I can change the boosts and you can try again
-
reosarevok
heh
-
-
church paris returns paris*h* churches
-
samj1912
lol XD
-
that's prolly due to ngarm
-
reosarevok: should I do any address boosts?
-
reosarevok
I'd say no, at least for now
-
Did the current search boost it?
-
samj1912
I dont know
-
I dont think the current search does any of its boosting the way we are currently doing it
-
reosarevok
Oh ok
-
samj1912
reosarevok: done
-
weirder results
-
I guess I need to boost area
-
reosarevok
Back in 5
-
samj1912
yo
-
reosarevok
wtf :D
-
The first doesn't even have Paris?
-
(but area is Church)
-
I think area + address might be too strong
-
samj1912
okay
-
reosarevok: try now?
-
reosarevok
The first is Paris Theatre... which is great, except it has nothing to do with a church :D
-
Let me try London instead, to be using name more than alias
-
hah
-
-
Well top 1 looks about right :D
-
samj1912
lol :P
-
reosarevok
But I would expect St James's Church, Piccadilly, London (which even has both words in the name!) to score higher than "London Coliseum"
-
samj1912
well its got the same score
-
reosarevok
Sure, but why, if both terms match? Shouldn't that matter? :/
-
samj1912
hmm
-
let me try and boost that
-
reosarevok: the problem with that is that st. james doesnt have any alias
-
other docs have aliases that are getting boosted I think
-
reosarevok
-
And both have London in the address and the area
-
samj1912
the problem is that they are given the same score by solr
-
which is weird
-
reosarevok
yeah
-
-
samj1912
well, that makes sense, since its edismax
-
the main problem is phrase search
-
hmm
-
ruaok
iliekcomputers: can we log a user in for 5 minutes with no automated re-logins?