reosarevok: we have a sponsor button on our repos that links to MeB.org, this is github sponsors, which people might find easier to use. For example: MeB.org only takes cards or PayPal.
2020-05-13 13434, 2020
reosarevok
Oh
2020-05-13 13453, 2020
reosarevok
What other things are there :D Bitcoin?
2020-05-13 13458, 2020
reosarevok
Anyway, sure, whatever is easier
2020-05-13 13402, 2020
reosarevok
As long as it keeps being zero fees, guess there's no reason not to
Oh. I mean, we have that too but I never expected anything international to support it
2020-05-13 13417, 2020
reosarevok
But I guess this can differ per-country :)
2020-05-13 13433, 2020
reosarevok
Cool, if they do support that!
2020-05-13 13438, 2020
reosarevok
(at least for countries bigger than Estonia :p )
2020-05-13 13420, 2020
ZaphodBeeblebrox joined the channel
2020-05-13 13420, 2020
ZaphodBeeblebrox has quit
2020-05-13 13420, 2020
ZaphodBeeblebrox joined the channel
2020-05-13 13451, 2020
CatQuest has quit
2020-05-13 13401, 2020
Mr_Monkey
Moin!
2020-05-13 13451, 2020
jmp_music
Morning!
2020-05-13 13449, 2020
BrainzGit
[listenbrainz-server] vansika opened pull request #845 (vansika/candidate-recordings…mapping-matchable-re): Mapping matchable re (to undo force commit) https://github.com/metabrainz/listenbrainz-server…
2020-05-13 13420, 2020
ruaok
mooion!
2020-05-13 13403, 2020
diru1100
Mo''ing!
2020-05-13 13400, 2020
diru1100
yvanzo: Hi, I am trying to create synthetic data using scikit learn. I will take excel sheet given by Freso, the plots done so far as reference and write the data generation code. Will update you once done.
2020-05-13 13411, 2020
BrainzGit
[listenbrainz-server] vansika merged pull request #845 (vansika/candidate-recordings…mapping-matchable-re): Mapping matchable re (to undo force commit) https://github.com/metabrainz/listenbrainz-server…
2020-05-13 13429, 2020
v6lur joined the channel
2020-05-13 13451, 2020
yvanzo
diru1100: ok, thanks
2020-05-13 13402, 2020
Gazooo has quit
2020-05-13 13418, 2020
jmp_music has quit
2020-05-13 13445, 2020
Gazooo joined the channel
2020-05-13 13424, 2020
jmp_music joined the channel
2020-05-13 13402, 2020
cloverfield has quit
2020-05-13 13420, 2020
BrainzGit
[listenbrainz-server] vansika opened pull request #846 (vansika/candidate-recordings…candidate-recordings-tar): Create periodic candidate recordings tar for recommendation toolkit https://github.com/metabrainz/listenbrainz-server…
2020-05-13 13441, 2020
D4RK-PH0ENiX has quit
2020-05-13 13420, 2020
D4RK-PH0ENiX joined the channel
2020-05-13 13432, 2020
Mr_Monkey
ruaok: Hi! I did an evaluation of the playlist capabilities. I took the 'recent' page's listens and made a playlist out of that. Out of 50 tracks I was unable to play only 5, two of which I could have found on spotify if I omit the release name (ie make the search less specific). Another two would have been the first result of a youtube search.
2020-05-13 13445, 2020
Mr_Monkey
One unplayable track out of 50 sounds pretty good.
2020-05-13 13414, 2020
ruaok
that certainly sounds like a very good starting point.
2020-05-13 13426, 2020
Mr_Monkey
I've got a few things to fix on the player, so I'll continue that, but I think after making the search less precise we'll be in a good spot.
2020-05-13 13431, 2020
ruaok
is the "make search less specific" by omitting the release name automatic?
2020-05-13 13454, 2020
Mr_Monkey
I was thinking of just ignoring release names entirely
2020-05-13 13400, 2020
Mr_Monkey
(For search)
2020-05-13 13424, 2020
ruaok
yes, that is what I am doing in the recommendation toolkit
2020-05-13 13437, 2020
ruaok
at this level it's just not relevant -- at least for now.
2020-05-13 13442, 2020
Mr_Monkey
Agreed.
2020-05-13 13407, 2020
ruaok
so, lets see if we can follow iliekcomputers' urging and see if we can get the playlist feature on track to production soon.
2020-05-13 13441, 2020
ruaok
which will require some documentation and other bits of polish.
2020-05-13 13453, 2020
Mr_Monkey
The problem in this case is more with bad tagging of the listens (the album names were borked), so potentially less of a problem with data coming from MB. Regardless, IMO if it spits out a song with the right name by the right artist it's good enough
2020-05-13 13416, 2020
iliekcomputers
i'm happy to keep tickets for documentation of endpoints while they're still being worked on.
2020-05-13 13443, 2020
iliekcomputers
maybe ishaanshah[m] or shivam-kapila can pick them up later
2020-05-13 13449, 2020
Mr_Monkey
Speaking of: are we OK with the /playlist endpoint? I was also thinking of /player
2020-05-13 13402, 2020
ruaok
Mr_Monkey: that is something we can tweak later, for sure. but having a playlist enabled LB deployed would be super useful.
2020-05-13 13415, 2020
ruaok
+1 /player
2020-05-13 13453, 2020
Mr_Monkey
OK, I can have a look at polishing, but I'm sure I'll hit my python wall pretty soon.
2020-05-13 13404, 2020
Mr_Monkey
At least I can catch a few obvious possible errors
2020-05-13 13423, 2020
ruaok
there isnt much python code is there? the view looked fine for me.
2020-05-13 13439, 2020
ruaok
don't worry about python bits for now -- I'm happy to help cover those.
> maybe ishaanshah[m] or shivam-kapila can pick them up later
2020-05-13 13431, 2020
shivam-kapila
iliekcomputers: Sure thing
2020-05-13 13405, 2020
shivam-kapila
ruaok: Did you get chance to look into the gist?
2020-05-13 13426, 2020
jmp_music has quit
2020-05-13 13418, 2020
ruaok
shivam-kapila: I did. I'm still thinking about it....
2020-05-13 13433, 2020
ruaok
how did you get those timings? with explain analyze?
2020-05-13 13423, 2020
shivam-kapila
No those timings are the execution timing for code in /<user_name> route
2020-05-13 13429, 2020
ruaok
because those times are still too long. but I don't have enough context about them.
2020-05-13 13435, 2020
ruaok
shivam-kapila: ok, that might not be the best way to test that -- for comparison of the various queries, ONLY go by explain analyze. If you don't there are other factors baked in that might affect the timing.
2020-05-13 13422, 2020
ruaok
however, one thing we should do is check to make sure that the DB lookup is the think that takes up the most time for this request -- if there is anything else taking time, we need to look at it.
2020-05-13 13405, 2020
shivam-kapila
The DB lookup takes the most time
2020-05-13 13400, 2020
shivam-kapila
I didnt use EXPLAIN ANALYSE for the same reason to check how much time it takes for full route processing and the DB lookup.
2020-05-13 13426, 2020
shivam-kapila
The other parts in the route take around 1 sec only
2020-05-13 13437, 2020
shivam-kapila
DB lookup sometimes jumps to 4sec
2020-05-13 13458, 2020
shivam-kapila
I will add explain analyse observations too
2020-05-13 13444, 2020
ruaok
all of that is very very slow, sadly. how fast is your laptop?
The bound makes it open only 4 hypertable chunks. I had set a limit to 3 weeks
2020-05-13 13443, 2020
ruaok
I'm on the timescale slack to ask about this too
2020-05-13 13431, 2020
shivam-kapila
Hm. Did you get any info from the team?
2020-05-13 13442, 2020
ruaok
I just asked -- no responses yet
2020-05-13 13434, 2020
shivam-kapila
A bound is good to go for. But somehow we also ned to see all the things going in the route. Its basically to much processing in there IMO
2020-05-13 13447, 2020
ruaok
yeah.
2020-05-13 13429, 2020
ishaanshah[m]
alastairp: Hi
2020-05-13 13451, 2020
ishaanshah[m]
I have setup eslint and prettier for AB
2020-05-13 13425, 2020
ishaanshah[m]
there are ~1700 problems, ~1400 of which are auto fixable
2020-05-13 13429, 2020
ishaanshah[m]
I will add eslint-disable at the top of each file for now, so the the github action wouldn't complain
2020-05-13 13450, 2020
ishaanshah[m]
Does that sound fine
2020-05-13 13415, 2020
alastairp
hi ishaanshah[m], thanks for setting that up. Can you give me an example of the kinds of errors? it seems like there's an error per line... is this mostly stylistic errors, or are there code improvements to be made too?
2020-05-13 13447, 2020
ishaanshah[m]
Most of them are stylistic errors
2020-05-13 13450, 2020
alastairp
there are only 1440 lines of javascript in the whole project
2020-05-13 13454, 2020
ishaanshah[m]
Which will be auto fixed
2020-05-13 13458, 2020
alastairp
so it's basically saying that every single line is wrong
2020-05-13 13416, 2020
alastairp
can you give me an example of some of these stylistic errors?
2020-05-13 13434, 2020
ishaanshah[m]
Default indentation for prettier is 2 spaces
2020-05-13 13443, 2020
ishaanshah[m]
Whereas the project uses 4
2020-05-13 13402, 2020
ishaanshah[m]
So that makes up for almost all of them
2020-05-13 13433, 2020
alastairp
can we tell prettier to use 4?
2020-05-13 13445, 2020
ishaanshah[m]
Yes we can
2020-05-13 13455, 2020
ishaanshah[m]
Should I do that?
2020-05-13 13418, 2020
alastairp
let me think about it. I'm not 100% convinced that it's worth changing _every single line_ in the project just to switch from 4 spaces to 2
2020-05-13 13457, 2020
ishaanshah[m]
Also the js uses both commonJS and import/export
2020-05-13 13431, 2020
ishaanshah[m]
I have configured it to use import/export,
2020-05-13 13443, 2020
ishaanshah[m]
So those make up for some more
2020-05-13 13401, 2020
ishaanshah[m]
Other than that a lot of var -> const/let
2020-05-13 13445, 2020
alastairp
yes, I'm happy with those two changes. that's one thing that I wanted to improve
2020-05-13 13418, 2020
alastairp
let's do that then. Can you change to import/export, and var->const/let, and configure the spacing as 4
2020-05-13 13423, 2020
ishaanshah[m]
Other than the indentation, most of the issues I see are because of usage of old javascript which makes sense as they were written a long time ago
2020-05-13 13424, 2020
alastairp
and then open a PR and we'll see what else is remaining
you'll have to add the workflow file from https://github.com/metabrainz/acousticbrainz-serv…, just copy the file directly (and make any changes you need to make), we'll close this PR without merging it
