we actually need to run the models. I should do that
also, we're waiting on my code for the actual evaluation, so we won't know results until that's done
ruaok
that's really cool. it is nice to have a vision become reality.
not bad progress for under 2 years of progress. :)
D4RK-PH0ENiX has quit
D4RK-PH0ENiX joined the channel
alastairp
yeah. there's a small handful of things that need to come together to make this easier to use
but it's definitely happening
I really want to get search going, really badly
ruaok
now for nearly 2 years. :)
alastairp
and Gentlecat wants to get feedback from the website going - "This feature is wrong, it should be x"
sigh
yeah
ruaok
that is really necessary, yes.
I can't wait for 6-12 months past these things working when the data shows concrete improvements
alastairp
unfortunately we still have no phd students working on it
that's my real dream
well, anyone, really
ruaok
there was one, yes? that person done, now?
Freso
ruaok: Speaking of you being away, I replied to your "mattress" mail, but didn't get a reply back.
alastairp
well, his work isn't "extend AB to do x"
it's "do this thing, which can be integrated into AB"
so I'll push to make sure the integration becomes a reality, but...
ruaok
ah.
D4RK-PH0ENiX has quit
Freso: meh. the dsb.dk site is broken in chrome
D4RK-PH0ENiX joined the channel
D4RK-PH0ENiX has quit
D4RK-PH0ENiX joined the channel
Freso: my plan is roughly, this: Arrive at 15:30 at CPH, take the train to København H, hopefully meet you there, possibly shop for a few things, then take the train to Rødvig
JesseW joined the channel
diana_olhovyk has quit
armalcolite: I'm trying the api_compat branch on commit 5b09 again and I'm still getting the 500 error: BadRequest: JSON document must contain the key listened_at at the top level.
I thought that was fixed.
mihaitish has quit
kartikgupta0909 has quit
armalcolite: it also occurs to me that I have no idea how to start using the api_compat branch. :)
you ok meeting me at københavn H with the mattress?
Freso
Sure. :) Valby would obviously be best for me (I live less than 1 km away), but Cph H is definitely doable and workable too.
Oh, and also re: a comment on your review at a meeting some weeks ago: "Each organization receives $500 USD per accepted student no matter whether the student passes or fails." So the two students we have had drop out should not impact what the org. gets from Google for money, AFAICT.
ruaok
I was more referring to the student money going unused.
if a student drops out that is $5k that wasn't pulled out of Google's pocket. :)
Freso
Ah. Right.
mihaitish joined the channel
Gentlecat
ok, mbspotify seems to be working with docker too now
alastairp
Gentlecat: I'm going to add a datasets section to /admin
for now it'll show the same stuff as the dataset list page, and have 3 options: mark from running->pedning, mark running->failed, promote finished to evaluation status
promoting to evaluation will compute highlevel against them and store in a temporary highlevel table
from this, we can do the cross-collection evaluation which you need
but I have a better way to do it, it shouldn't be much work to make the change from how it is currently
the difference is that I wanted to create a new temporary highlevel table per model to evaluate
now that we have the `model` table and `model.status`, we can add a new model based off an evaluation job, and keep track of the different highlevel files
I'll try and write it up in a ticket for you to give a look at it, if you want
I wonder if it makes sense to be able to first compute the lowlevel files that are a part of the evaluation set
that way we can do the evaluation more quickly, instead of having to wait for all 3 million to finish
Gentlecat: I got a really basic 'migrate' command working, which finds unapplied update files and runs them
currently only works for sql files, not python, and doesn't recover from sql errors
but seems OK for now
Slurpee has quit
rahulr
bitmap: Ping.
bitmap
pong
rahulr
Hi :)
bitmap
oh, friday again. hi!
sorry I hadn't checked the code in a few days until now
rahulr
Oh, that's fine.
Have a look whenever you find time. Though I've started working on the other dialog, on a separate branch.
bitmap
I'll try to edit some stuff locally and see if I can fix that whitespace issue
good
the track one is starting to look fairly complete to me
I'll cc some other picard devs on the PR and see if they can take a look
rahulr
I was thinking of reusing the `release_to_metadata` method from mbxml.py (https://github.com/metabrainz/picard/blob/maste...), so as to avoid writing a parsing method, but couldn't make it work till now.
Yeah. Sure.
bitmap
ah, good idea, I hadn't thought of that
you have it working now though?
rahulr
Nope. Not for release. I guess it wasn't meant for a release node from a release query. Rather for a release lookup through mbid. I am getting some attribute errors. But I think I can fix that.
I'll have to look more into it.
bitmap
would it help to reuse those *_to_metadata functions in parse_tracks_from_xml?
our XML isn't very consistent between lookups and searches
ideally we have standard parsing code used everywhere but it might be tricky
rahulr
Yes, it can be reused for tracks. But most of the information extracted is wasted. I'll try using it and see if it doesn't make much of a mess.
bitmap
that's fine, it's not storing a lot of results in memory
rahulr
Ok. I'll try that. See if I don't break things.
Another thing...
Should I be saving dialog state on pressing cancel (this includes advance query syntax, window size, and column sizes). As the "Load into Picard" button is disabled until an item is selected.
bitmap
I would have the window and column sizes always persist (even across different tracks)
the advanced query syntax should persist anyway since it's using/modifying the actual setting, right?
rahulr
Not if it's state is changed in dialog. If the user clicks cancel, it'll return to original state.
hecanjog joined the channel
hecanjog
Hi all!
I work for the freemusic archive, we're really interested in checking out acousticbrainz in the FMA, just wanted to say hello
bitmap
I guess the settings change should be reverted, since that's how the options dialog works
I would have the window/column sizes persist though since it doesn't make sense to cancel that
rahulr
Umm. Ok. I'll do that.
bitmap
the cancel button is canceling the form, not how it's displayed
people will resize things to how they like and expect that to stick