bitmap: if the authorization header is set to empty that is also considered inavlid access token
bitmap
that has scripts/edit.js: "scripts/edit-4da2c51e9d.js"
so it knows the edit bundle will be named like scripts/edit-4da2c51e9d.js at the public url
xarph has quit
and it's a hash of the file contents, so if the bundle changes, so will the hash (and clients will download the new copy instead of the old, since it has a different name)
spellew
ferbncode: iliekcomputers: I've opened the draft pr
bitmap
amCap1712: it also returns a 401, but with a different message ("You are not authorized to access this resource.")
"<pristine__> any user must have listened to a song that involves various artists, we can then get similar artists from the histogram project and recommend songs from a pool which belong to these related artists" I think the VA based histogram project is very cool, but I also feel like this approach (of using VA albums as a base for artist similarity) may exclude a lot of smaller and newer or just straight up less mainstream (with
"mainstream" meaning "mainstream in Western mostly-English locales") artists. Both based on bias of more mainstrain = on more compilations (incl. on compilations at all!) and also of the bias in the data available/the historical userbase. It also excludes artists that aren’t following the typical Release workflow, but only does standalone recordings on SoundCloud or YouTube etc. I'd rather get recommended smaller, less mainstreamy
artists than artists that happen to have been on dozens of compilations.
Anyway, I also realize that this is a difficult problem, so I’m not saying to drop this approach; I just really hope all the downfalls of this approach won’t be forgotten once there is something Good Enough™ in place. :) (And I’m sure you and Rob have thought about these limitations, I just didn’t see either of you express them here in the channel.)
ZaphodBeeblebrox
oh definitly.
I'm thinking about Mash-upers like who have a lot of standalone recordings