oh, whoops. sorry about that. forgot to remove it. the class itself is ready for review.
ruaok
ok
amCap1712
gr0uch0mars: that Google library we can't use because it is not compatible with fdroid
I had asked in their IRC channel they asked me to use an alternative
gr0uch0mars
oh I see :(
Cyna
bitmap: I've pushed made the changes and pushed the code for AddRemoveAlias.js please check it when you have time. I'll raise a few more PRs once this is merged
ruaok: do not merge that one yet, I'll run it past the two of them again in case they want changes to the text since it's been a while since I wrote it :)
travis-ci joined the channel
travis-ci
metabrainz/picard#4688 (master - 6940062 : Laurent Monin): The build passed.
alastairp: hi, can we touch base about the main project? I want to create a PR today to introduce the similarity metrics to AB - do you think it's a good idea to make the necessary migrations now and compute the metrics incrementally using a cli script? Once we have computed the metrics I am ready to begin integrating annoy with api endpoints and a public evaluation.
ruaok
I really didn't think that someone other than me would have the first THC reference on meb.org lol
ruaok: don't deploy until they say it's 100% ok please :)
ruaok
oh.
ruaok stops
reosarevok
(I said it earlier! :P)
ruaok
where?
reosarevok
" ruaok: do not merge that one yet, I'll run it past the two of them again in case they want changes to the text since it's been a while since I wrote it :) "
here! :p
They've both seen the text etc but I sent a screenshot of how it looks + the pics now for a last check
ruaok
oookay.
iliekcomputers
>When not looking all over the internet (and the library!) for instrument data, other hobbies include playing games, listening to all sorts of different music, Lego, colouring, and collecting all sorts of things, from stickers and stamps to bottlecaps.
this sentence seems incomplete?
ruaok
seems fitting to me.
iliekcomputers
oh yeah, i misread the "Other hobbies include" part
ruaok: re the artist popularity accuracy, we could have a page on MB where people could vote on whether the MBID seems accurate or not or something like that.
ruaok
how do you go about finding the questionable cases?
iliekcomputers
i mean, a webpage with an (Artist name, Listen Count, MusicBrainz link, some other artist data) table would probably show some of the major cases
alastairp
aidanlw17: hi, I've got some things to finish today, how about around this time tomorrow instead?
ruaok
iliekcomputers: there would be thousands of pages to review....
reosarevok: more seriously, yes. but we need to see the actual data first and paw through it.
aidanlw17
alastairp: ok I can make it today anyways and just adjust if we decide differently tomorrow
alastairp
but yes, an incremental update script would be good for the metrics
yeah, in fact a PR is the right place to have this discussion
so, I understand that you can already create an annoy index, and query it
will that be included in the PR?
aidanlw17
That’s correct, I can
I could include those things
alastairp
you should
aidanlw17
Okay - I thought it might be better to split them into separate PRs as separate tasks?
I can include those things as well though
alastairp
your evaluation will be based on what I can see, not what you've told me exists - a full end-to-end pipeline that 70% works is better than you just telling me that it works :)
ruaok
+100 to that alastairp
alastairp
mm, right. what does this entail then - do you have this functionality in separate commits/branches?
if you were planning on splitting it into PRs, do you know how you'll commit the separate functionality?
aidanlw17
It’s all in one branch that I was building but to make the PRs I was going to move the different parts into separate branches
E.g. I can submit the entire pipeline today, but over a course of ~3 PRs
1. Compute similarity metrics
2. Add annoy class and compute annoy indices
3. API to query annoy
alastairp
if it's easy for you to make separate PRs, do that. but don't spend extra time trying to split it up if it's more difficult
but yes, get these PRs out as soon as you can! I want to run it myself
aidanlw17
Okay I don’t think it will be too difficult to do it in that way
I’ll make them today!
One question I had is I’m not sure where we should be saving the static annoy indices when we move this code over to the AcousticBrainz server? This might be a little bit of my confusion with docker but is it best for us to save them to disk inside the rest of the codebase?
alastairp
don't worry about that, we can discuss it in the PR
Okay sounds good - for now they are just being saved in a directory with the code so that will need to be changed
alastairp
don't worry about specifics for deploying it, that's up to iliekcomputers to fix
aidanlw17
Oh okay I see! That makes more sense. Thanks
Should I try to change any of that or just leave it as is for the PR?
alastairp
I don't mind -
if you want to try and learn how to do it, go for it
iliekcomputers
just make it easy to specify the path, config setting would be ideal
alastairp
if not, I'm going to tell you how to do it in the PR
remember that the PR doesn't have to be perfect for you to submit it. it's fine even to open it when you haven't finished coding, so that we can give feedback on the design
aidanlw17
Got it - with that in mind I probably should have opened a PR for this earlier