if I wasn't looking at the docs I'd just expect this to return either summary of recordings or all data that we have on them
2016-10-03 27749, 2016
alastairp
GET /api/v1/low-level?ids=
2016-10-03 27750, 2016
alastairp
perhaps
2016-10-03 27710, 2016
alastairp
similar to the existing GET /api/v1/(uuid: mbid)/low-level
2016-10-03 27711, 2016
Gentlecat
recording_ids
2016-10-03 27716, 2016
alastairp
better
2016-10-03 27723, 2016
Mineo
the list of open pull requests for picard at https://github.com/metabrainz/picard/pulls has been slowly (but unfortunately steadily ☹ ) growing, with a lot of pull requests tagged "Decision Required" with just 1 or 2 persons giving feedback. if anyone of you (even as a user, not developer of picard) could spare a few minutes to look at them, that would be great!
2016-10-03 27735, 2016
Gentlecat
I'd still love to be able to reference each submission
2016-10-03 27740, 2016
Gentlecat
without an offset
2016-10-03 27748, 2016
alastairp
we *do* have row ids
2016-10-03 27752, 2016
Gentlecat
pretty sure it would be useful for dataset stuff too
2016-10-03 27754, 2016
alastairp
what's your usecase?
2016-10-03 27701, 2016
Gentlecat
are they UUIDs?
2016-10-03 27704, 2016
alastairp
no
2016-10-03 27707, 2016
alastairp
just ints
2016-10-03 27729, 2016
alastairp
in datasets I do think it would be a good idea to use row ids in some places
2016-10-03 27729, 2016
Gentlecat
for example using a specific submission to train a model
2016-10-03 27740, 2016
alastairp
I see it like this:
2016-10-03 27704, 2016
alastairp
a dataset contains mbids, and a filter function, to choose a specific duplicate
2016-10-03 27726, 2016
alastairp
which could be first submitted, first flac, etc
2016-10-03 27741, 2016
alastairp
apply mbid + rule -> ids -> snapshot
2016-10-03 27748, 2016
alastairp
so that snapshot always stays the same
2016-10-03 27700, 2016
Gentlecat
right
2016-10-03 27704, 2016
alastairp
but applying the same rules to the same dataset later may result in a different snapshot
2016-10-03 27728, 2016
Gentlecat
so I guess what I have in mind is one of those rules
2016-10-03 27747, 2016
Gentlecat
use a specific submission for this recording
2016-10-03 27700, 2016
Gentlecat
if I want to build a model just with my own data
2016-10-03 27703, 2016
alastairp
they're definitely important. I'd like to see a rule which pre-filters submissions to remove obviously incorrect songs
2016-10-03 27705, 2016
Gentlecat
what do you mean by incorrect?
2016-10-03 27707, 2016
alastairp
right
2016-10-03 27728, 2016
alastairp
someone may have submitted a track with the wrong mbid
Freso sitting in the office now with ruaok, Gentlecat, and bitmap
2016-10-03 27707, 2016
chirlu
UmkaDK: Yes, there are a few historical things that aren’t actually needed anymore. I personally know of our own median implementation that can be done directly in Pg 9.5+.
2016-10-03 27729, 2016
chirlu
Whether other options for unaccent/collate do exactly what is needed would need closer review, of course.
2016-10-03 27720, 2016
UmkaDK
sure! I'll add the code to the ticket (when I raise it) and leave it in your hands guys. :)
2016-10-03 27758, 2016
Freso
UmkaDK: Thanks. :)
2016-10-03 27751, 2016
Freso
UmkaDK: I don't know how closely you've followed the channel, but ruaok set a hard deadline for the switch in the beginning of November, so we'll Soon™ be able to look into things like that.
2016-10-03 27724, 2016
ruaok
I just booked a car too. that car that will haul all of our precious servers to the recycling station ;_;
2016-10-03 27701, 2016
UmkaDK
No worries! It's just something we came across in the process of moving to the new infrastructure and were wondering if we could cut a corner there.
