Thanks ruaok. Not surprising to see and Amazon baby closing up access to information, but always disappointing.
ruaok
yep. more fodder for BB.
Gazooo79494 has quit
Gazooo79494 joined the channel
xuvxuv joined the channel
alastairp
morning
I'm at a conference today/tomorrow so won't be around much until wednesday
I should be available for reviews if needed
ruaok
I have no PRs that are in need of review -- which means its time to fix that. :)
alastairp: I'm thinking of the mechanics of creating end-of-the-year review playlists. I though about deploying them on test, but right now we cannot create a playlist for a user who doesn't exisit. and since almost no one has logged into test, the target demographic is... urm. effectively zero.
we could relax the requirement that a createdfor playlist must be created for a user who doesn't exist -- it might just be a validation step.
then I could create playlists for all LB users on test.
I've started doc to think about what could go into "year in review" playlists. if you have ideas, please leave a comment.
alastairp: Mr_Monkey ^^
alastairp
ruaok: what do you mean about relaxing the requirement? currently we require that `created_for_id` exists in the user table (because we link with an id, rather than a textual name)
are you suggesting that we change it to a name, and then change the checks?
alternatively, 1) copy user table from prod to test, 2) release in beta but disable the endpoints in both prod and beta (wouldn't have endpoints to add playlists, we can still run stuff from commandline)
ruaok
hello, can of worms.
I'd rather not do any of that. if it was relaxing a validation check, then fine.
now I am more in favor of us trying to get the playlist stuff somewhat sane this wednesday with the goal of releasing it into production early next week. with a big fat BETA label on the playlist features, but still running on prod.
Looks like someone released to their auth without running tests
sumedh has quit
adhi001 joined the channel
sumedh joined the channel
davic joined the channel
_lucifer
shivam-kapila: they actually can do that, google has a mono repo so all tests are automatically for all changes. if they are consistent failures, the change is automatically roll backed
shivam-kapila
Interesting
sumedh has quit
sumedh joined the channel
v6lur joined the channel
D4RK-PH0ENiX joined the channel
D4RK-PH0_ has quit
ruaok
one query and bam, another 17M row dataset. lol.
sumedh has quit
D4RK-PH0_ joined the channel
D4RK-PH0ENiX has quit
sumedh joined the channel
angryhippy joined the channel
r0han22 joined the channel
Gore has quit
rohan_ joined the channel
thomasross joined the channel
Mr_Monkey
ruaok: Looking at the playlist editing, I'm getting 400 errors because "JSPF playlist public field must contain 'true' or 'false'.".
Looking at the serialized JSON data I send, the 'public' values are booleans rather than strings (`true` vs. `"true"`) which is valid JSON (generated by `JSON.stringify`).
I suppose both should be allowed, but if we must choose one bools would be better IMO.
ruaok
oh, duh.
alastairp: looking up the first listened data I've been working on and doing typsense mbid mapping lookups, here are the first attempt stats: 636 good matches, 364 bad matches. 64% good matches
thats at least 10% better than the previous mapping.
every single track of !!! fails, of course. lol.
alastairp
cool, good to see that it's better than the previous one!
ruaok
and there are stupid errors to fix, as well. the index is case sensitive right now!
People up for reviews: CatQuest, ruaok, yvanzo, zas, reosarevok, alastairp, Mr_Monkey, bitmap, Freso, shivam-kapila, _lucifer, jmp_music – anyone else who want to give review, let me know ASAP!
No mailed-in reviews this week, so…
CatQuest: Go!
(Song provided courtesy of the resident Mr_Monkey.)