Lucifer: my favourite Onion feature!! (regarding the tweet though, the headline for those is always real news, the comments that come with it is the satire)
mayhem: interesting question. I'd say that bits of messybrainz are probably the same, the flask app structure, bits of the ingestion API, and the base flask templates, probably not much more left from that
mayhem: of course, knowing academia, having more than 1 article that reproduces the same work is definitely a good idea!
2022-01-27 02727, 2022
mayhem
well, its not so much about anything concrete in that paper, but there is a seed thought that turned out to be rather insightful.
2022-01-27 02728, 2022
alastairp
actually - speaking of that article, I wanted to run some stuff past you today. work out where we should try and get to in the next week or so
2022-01-27 02739, 2022
mayhem
I'd love that.
2022-01-27 02747, 2022
mayhem
in the office today?
2022-01-27 02718, 2022
alastairp
I teach this afternoon, so it kind of breaks up my day a bit too much. I'll be in next week
2022-01-27 02743, 2022
mayhem
k.
2022-01-27 02717, 2022
mayhem
I'm working on a proof of concept of something, that if it does pan out, could be a bit of a game changer for us.
2022-01-27 02724, 2022
alastairp
great
2022-01-27 02741, 2022
mayhem
that would influence this discussion, possibly a lot.
2022-01-27 02747, 2022
mayhem
let me keep working for a bit.
2022-01-27 02703, 2022
mayhem
this of course was one of those 11pm insights and then you can't stop hacking until 2am...
2022-01-27 02724, 2022
alastairp
I think the BPM stuff, while useful, is probably pending on getting an interesting dataset of ground truth - that could be either related to us collecting data from users, or somewhere else
2022-01-27 02741, 2022
mayhem
as are many others yes.
2022-01-27 02712, 2022
alastairp
one options would be to go ahead and do our plan for running all algorithms, and then get some spotify ids for songs for which there are inconsistencies, and start putting them in front of people
2022-01-27 02713, 2022
mayhem
which makes me want to focus on BP as a killer experience, complete with data gathering tools
2022-01-27 02723, 2022
mayhem
and then build ML tools out.
2022-01-27 02747, 2022
alastairp
however, I'm not sure exactly what that's going to give us in the long run - as you said, getting candidate bad songs in front of software developers is a good idea. is this process enough?
2022-01-27 02715, 2022
mayhem
I guess the process is enough if it yields sufficiently clean data. which of course defies definition.
2022-01-27 02709, 2022
aerozol has quit
2022-01-27 02715, 2022
alastairp
yeah - and even on my small test of queen, pink floyd, and miles davis we saw stuff which I'd define as not clean data
2022-01-27 02711, 2022
mayhem
which in all of this makes me want to go back to our roots. have users give us this info.
2022-01-27 02725, 2022
alastairp
yeah, right
2022-01-27 02735, 2022
mayhem
what was this guy Bob's name who kept harping that MIR research wasn't any better than a horse could so or so?
2022-01-27 02744, 2022
mayhem
I think he's right. 100%.
2022-01-27 02751, 2022
alastairp
yep - Bob Sturm
2022-01-27 02759, 2022
mayhem
him.
2022-01-27 02713, 2022
alastairp
right, his argument was to try and think about things in terms of musical understanding
2022-01-27 02743, 2022
alastairp
e.g., want to classify jazz? start looking at instrumentation and chord sequences
2022-01-27 02706, 2022
alastairp
of course, that's a difficult task in itself, so...
2022-01-27 02722, 2022
mayhem
I'm really starting to think that these hard problems "what is jazz" are just pointless to solve. very hard, poor results.
2022-01-27 02702, 2022
mayhem
I really think that given a goal in mind (e.g. we want to do recommendations and cross-platform playback) we can find simple solutions that may not even require machine learning. or much fancy stuff at all.
It was mainly for the playstore release process of the app because the process requires us to do 1-2 additional things, which can be missed by someone. But I agree with the point that since we are a few, we don't have to worry about this. However, I will make this point for future devs in the docs in case we expand and multiple people are involved in the process!
2022-01-27 02756, 2022
monkey
👍
2022-01-27 02720, 2022
alastairp
in that case, clear documentation is definitely an important part of the release process
2022-01-27 02736, 2022
alastairp
and is exactly why we just spent a week doing this, because there are many parts for which only one of us knew how to do something
2022-01-27 02749, 2022
akshaaatt
Yusss 💯
2022-01-27 02755, 2022
monkey
I'll add that in some cases it's useful to allow the team access so someone can do a release for you if you're unavailable (provided there's good docs on how to do it)
but do you think there is any way I could have bono to myself for another... half hour to an hour?
2022-01-27 02733, 2022
lucifer
sure 👍
2022-01-27 02736, 2022
mayhem
thanks!!
2022-01-27 02704, 2022
mayhem
python script at 80% of ram, lol. she's gonna blow capn'!
2022-01-27 02745, 2022
lucifer
lol
2022-01-27 02708, 2022
reosarevok
yvanzo, bitmap, lucifer, mayhem: zas asked us to document different bits of docker-server-configs which were not added by him, maybe you can take a look when you have the time :)
2022-01-27 02725, 2022
mayhem
is there a PR for this project yet or should we start one?
2022-01-27 02708, 2022
outsidecontext
mayhem: I love the Bad Gateway sign. It's for sure a quiet place 😜
2022-01-27 02750, 2022
mayhem
yeah. peaceful.
2022-01-27 02739, 2022
reosarevok
mayhem: nothing yet, I think, because zas just added a basic readme directly for the stuff he knew
