16:17 PM
v6lur
16:17 PM
i changed my vote from yes to abstain, so it will fail otherwise
16:17 PM
hawke__
dswoe: Well, if the ripper stores a .cue file from the disc, then it’s possible.
16:18 PM
dswoe
yeah, strictly secret source.
16:18 PM
yep yep
16:18 PM
reosarevok wonders whether all the pages nested under Next Generation Schema http://wiki.musicbrainz.org/Special:Allpages&from=Next%20Generation%20Schema should be moved to History or to normal (not under NGS) pages
16:19 PM
weird design, that's all.
16:19 PM
hawke__
dswoe: Another way to do what?
16:19 PM
ruaok
reosarevok: please keep Migration_Plan.
16:19 PM
dswoe
find an album with a set of track lengths
16:19 PM
ruaok
I dont so much care of the other sub-pages.
16:19 PM
hawke__
Hmm.
16:20 PM
ruaok
dswoe: you could get a magnifying glass and read the CD-TEXT bits off the CD-ROM.
16:20 PM
thats another way.
16:20 PM
dswoe
balls.
16:20 PM
ruaok
16:20 PM
hawke__
dswoe: Well, the problem is that durations have resolutions in the second, whereas the discID is based on individual samples (1/75 sec)
16:21 PM
s/samples/sectors/
16:22 PM
dswoe
oldschool shit.
16:22 PM
16:22 PM
which is kind of what I want, but backwards.
16:23 PM
hawke__
well, it is a *disc* ID. :-)
16:23 PM
the more modern way of doing such things is with acoustID and such, but that doesn’t work with an entire disc at a time, obviously.
16:24 PM
Maybe we need tracklistIDs or something.
16:24 PM
Though that would have a lot more collisions
16:24 PM
dswoe
that would be the next heuristic after this step failed
16:25 PM
gunna write a strange story about my experiences here, then write perl to compile that into a music recogniser.
16:25 PM
it could take right off if it pirated shit for you.
16:25 PM
hawke__
heh
16:30 PM
dswoe_ joined the channel
16:31 PM
dswoe_
I'm back... "heh"
16:41 PM
ruaok joined the channel
16:53 PM
v6lur joined the channel
17:02 PM
kaiarneg joined the channel
17:28 PM
uptown joined the channel
17:35 PM
PasNox joined the channel
17:36 PM
luks
mb2freedb is actually just a "find an album with a set of track lengths" service
17:36 PM
I stopped using the freedb discids at all
17:37 PM
using the nearest tracklist search works much better
17:57 PM
hawke__
luks: So it just immediately converts the sector counts into durations?
17:58 PM
luks
yes
18:00 PM
hawke__
So theoretically, dswoe_ could just build a mbid by multiplying the durations he has by 75?
18:01 PM
or…wait, does that only apply to mb2freedb? What about the musicbrainz disc ID lookup web service?
18:02 PM
luks
the feature was written for the MB disc id lookup service, but never actually used :)
18:02 PM
hawke__
which feature?
18:02 PM
luks
the fuzzy tracklist lookups
18:03 PM
hawke__
So it won’t work for that?
18:03 PM
luks
no, it would work just fine
18:03 PM
he could simply create a fake TOC from track durations, send it to MB, and it would return the matching album
18:04 PM
I mean, it would work in theory, it's not implemented
18:04 PM
but that was one of the original design goals
18:04 PM
we even wanted to drop discids and use only this for CD lookups
18:05 PM
nikki
I thought ws/2 can do fuzzy lookups now? or are you talking about something else?
18:05 PM
hawke__
How do you accurately create a fake TOC though, given that each duration has 75 possible offsets?
18:05 PM
luks
it doesn't matter, the service will return you all tracklists where the durations are similar enough
18:05 PM
nikki: no, it can't
18:06 PM
nikki
no?
18:06 PM
nikki could have sworn she tried it
18:06 PM
ocharles
huh? sure it can
18:06 PM
ocharles notes he only joined this half way in
18:07 PM
luks
oh, well, I don't rememeber it ever being implemented
18:07 PM
ocharles
/ws/2/discid?toc=
18:07 PM
luks
well, that would work for dswoe_ then
18:07 PM
dswoe_
hmm
18:14 PM
hawke__
Looks like you need to provide a fake discid though, no?
18:16 PM
dswoe_
I find myself reading c
18:16 PM
dswoe_ throws it on the stack
18:16 PM
hawke__
I seem to get a lot of “invalid toc” too
18:17 PM
dswoe_
I can thump out some perl code for a more tracklistey ws/2 method
18:17 PM
I would love to, in fact
18:17 PM
scraping discogs now because sometimes lastfm doesn't bother having tracklists
18:18 PM
and I believe the sun is up outside
18:25 PM
hawke__
Ah, I see…I missed the toc format.
18:25 PM
Is that specified somewhere?
18:28 PM
ocharles
dswoe_: no, no need for a fake disc id
18:28 PM
I don't think anyway
18:29 PM
Mineo joined the channel
18:30 PM
hawke__
18:30 PM
18:30 PM
18:30 PM
(though of course it doesn’t find anything
18:32 PM
ocharles
It should still do a search on the toc if the disc id look up fails
18:32 PM
hawke__
I’m sure it does, that’s not a known-good toc.
18:33 PM
ocharles
oh, ok
18:33 PM
hawke__
just demonstrating that you do need to provide a (fake) discid
18:33 PM
ocharles nods
18:33 PM
ocharles
want to do the honours and open a ticket? :)
18:33 PM
hawke__
Sure
18:34 PM
Is the format for the toc parameter documented somewhere, btw?
18:34 PM
ocharles
i don't think so
18:34 PM
hawke__
Took me a bit to figure out that the second number is the tracks…
18:34 PM
ocharles
feel free to document that on the wiki, or give me a documentation ticket
18:34 PM
hawke__
Not even sure what the third number is
18:34 PM
I’d document it if I knew what it was. :-D
18:34 PM
I don’t know what the first 3 numbers are though.
18:35 PM
ocharles
my ($first_track, $last_track, $leadout_offset, @track_offsets);
18:35 PM
first track index, basically, and it's forced to be 1
18:35 PM
hawke__
uhhhhhh
18:35 PM
why even provide that then?
18:35 PM
ocharles
because.
18:35 PM
:P
18:35 PM
hawke__
:-D
18:35 PM
ocharles
(would do this now, but about to catch a train to watch the world burn. or perhaps just a bonfire)
18:36 PM
hawke__
A’ight.
18:38 PM
18:40 PM
ocharles
most excellent.
18:53 PM
dswoe_
cool
18:59 PM
hawke__
18:59 PM
Does it look right?
18:59 PM
ijabz joined the channel
19:09 PM
OK, so I’ve constructed a TOC from a digital release that I have, and the lookup doesn’t find it.
19:09 PM
19:10 PM
19:11 PM
19:17 PM
19:17 PM
which also doesn’t return anything
19:17 PM
So I guess the fuzzy matching doesn’t work?
19:19 PM
ijabz joined the channel
19:25 PM
dswoe_
are you multiplying seconds by 75? plenty of room for rounding errors there
19:26 PM
nikki
hawke_: I wanted to test whether fuzzy cd lookup only works with formats which have disc ids, but test won't let me submit the edits to find out
19:28 PM
hawke__
dswoe_: Yeah, but since I’m multiplying it by the tracklisting on the MB data, it shouldn’t have any rounding errors
19:28 PM
There were definitely some differences vs. the ones my music player showed.
19:28 PM
Anyway, it should be fuzzy enough for it to work anyway.
19:29 PM
dswoe_
oh yeah
19:29 PM
imo all this TOC bollocks is bollocks
19:29 PM
but I'm going sleepybyes right
19:29 PM
so good luck
19:29 PM
:)
19:33 PM
djce joined the channel
19:34 PM
hawke__
It probably would make more sense for the toc to be a list of track durations in seconds.
19:35 PM
That way an error/difference early on wouldn’t screw up the later ones
19:35 PM
nikki
19:35 PM
(it returns a whole load of other stuff too, unfortunately, since I'm too lazy to do loads of tracks :P)
19:38 PM
hawke__
which guess is that?
19:38 PM
nikki
that it only checks for mediums where the format can have a disc id
19:38 PM
hawke__
Oh, I see.
19:38 PM
*Format* which can have disc IDs
19:39 PM
I read it as “releases which have disc IDs”
19:39 PM
nikki
ah, nope
19:39 PM
hawke__
Good to know, though
19:40 PM
Think it’s worth filing a bug about?
19:41 PM
could always make a “version 2 toc” which would have the first number as 2, and then the rest be a list of track durations.
19:42 PM
(retroactively making that “first track index” into the version number)
19:42 PM
nikki
I don't think I would change /ws/2/discid, but I could see the use for something that works with for any format
19:42 PM
maybe /ws/2/release or something, no idea. I'm not responsible for the webservice :)
19:55 PM
praest76 joined the channel
19:57 PM
MJ joined the channel
19:57 PM
MJ
so, how do I best add a 2 CD release?
19:58 PM
MJ realizes it already.
19:58 PM
I have a 2CD album here, put first disc in MB Picard, lookup manually, add release with 2 discs, attach second disc id to same release.
