the hard thing is figuring out what really needs to be done.
people will scream for everything all the time.
Muzzz screams
luks
I've personally never seen a box set of a sets that is identifiable on it's own (e.g. by EAN or catalog number)
Muzzz
I meant to type 'creams' there
srotta
nikki: And yeah, I think that's ugly as hell, isn't really logical, and now we have multiple copies of AC/DC albums where the only difference is that one's named after a box set and the other one's not.
ruaok
luks: now the question is this: Can we come up with some sort of schema that uses our current schema and adds the BBC's CI while taking us a step towards NGS>
luks
ruaok: if I understand what they mean by CI, then yes
nikki is sure japanese box sets have their own catalogue numbers
ruaok
luks: would you do me a favor and sketch out how you see that could be done?
nikki
hmm... guess it depends exactly what you mean by it, I guess the box would have a barcode, the discs within the box set have catalogue numbers
ruaok
then we could take that fluff it up and toss it over to the BBC as a first stab at defining it and seeing if it fits their needs....
luks
nikki: yes, but you said you have a box set that includes box sets
warp
srotta: i think that's part of a larger problem, there are many other reasons to duplicate a release with the current schema. and _that_ is the problem.
luks
nikki: does the top level box set have a catalog number?
nikki: in my experience is just a package to sell more items at once
ruaok: ignoring artist changes, ignoring track changes
warp
luks: i think the brilliant classics Bach Edition qualifies.
ruaok
including the label changes?
nikki doesn't have all her CDs handy
luks
technically there are no label changes
nikki
well technically I do, I just don't have the key to open the box handy
ruaok
except for separating the label and labelname tables?
luks
yep
which isn't really that useful for labels
ruaok
so then medium would connect to our current album?
srotta
The whole box set thing is sort of a mystery to me, particularly the way it's documented and how it's implemented in edits, in practice. For example, me and inhouseuk have some differing views about an Elton John album (http://musicbrainz.org/show/edit/?editid=9432442)
warp
luks: the whole thing is one box with 155 cds in it, barcode and catalog. but the box has also been released in seperate volumes. Same music content, but the package uses a different volume/disc numbering scheme.
luks
ruaok: I'd like album AND tracklist to be the current album. album would take the meta data, tracklist would take the list of tracks and discids
ruaok: er, sorry, I meant from tracklist to right buttom
ruaok
ah.
and that would then essentially replace the album and release tables?
albumjoin would be where tracklisttrack is now?
luks
yes
ruaok
I think that is imminenty doable.
and I think the BBC will likely be ok with that.
luks
release technically stays untouched, except it has a title and multiple catalog numbers instead of one
ruaok
would it make sense to do albumtitle in this round or wait til next?
and the release.gid == summit8.PackageId ?
luks
every release needs a title, so either we have heavily duplicated data or we separate albumtitle from album
yes
ruaok
one go sounds smarter
ok, I think I grok.
thats a perfect sized chunk for an intermediary step.
and from the looks of it artistcredit might make another one.'
luks
I had one idea about how the *title/name tables are handled, but I want to benchmark it before
ruaok is curious
in my current schema they are bound to artist/album, the idea was to move the link to artist/album to artistcreditname and release
that would mean less text duplication in albumtitle, but a little more data in release
it would also mean a direct link from release to album
currently it can save ~100k album titles
ruaok
thats a significant change on the CI proposal, no?
how much effort to benchmark that?
luks
not really, I'd say it's just a technical detail
the core schema is still the same
ruaok
granted, but its a decision we need to make before we start on CI
luks
yes
ruaok
how will you benchmark that?
luks
implementing both in postgres
nikki leaves to get a train to uni
but I think for album/release it's quite easy decision
for artists it's harder
ruaok
since we can't quantify the data in the new structure.
well, since the core schema doesn't change, I don't really have to address this issue *now*.
luks
I'll try to draw some diagram how I'd integrate into the current schema
and some SQL how to convert the data
ruaok
awesome.
I'll start the conversation with the BBC to see if we can get this on the road.
Wizzcat
"321,271,968 4.74M/s eta 2m 45s" speedy pipe you got :>
luks
ruaok: still awake?
ruaok
barely.
luks
ok, let's talk tommorow :)
ruaok
Wizzcat: downloads off that machine are throttled.
I hope to fix that before too long, now that we have a better switch.
luks: will you be around in 8 hours or so?
luks
yes
ruaok
cool.
then again, I'm curious. again. :)
I can always re-read the logs in the morning, so hit me. :-)
luks
well, why is dateadded in table album? :)
it obviously should be in albummeta
ruaok
urm. ya.
ruaok makes a note
my GF will be out of town this weekend and there is rain forecast.
ruaok hopes for a nice coding weekend
luks has similar plans
warp
:D
ruaok
natta everyone...
ruaok has quit
ruaok joined the channel
so I managed to close my eyes...
and I figured out why dateadded is in album, not albummeta
luks
why?
ruaok
dateadded is not derivable from other datat. the moderation tables may not be present. or they may be pruned.
albummeta can be easily regenerated.
luks
hmm
ruaok
if you have any ideas, I'll read the chatlog in the am.
bad time.
natta
ruaok has quit
aCiD2 joined the channel
FauxFaux
aCiD2: Back at uni? \o/
UncleRemus has quit
Muzzz
His hostmask says yes :P
Muzzz is now at Birmingham New Street waitinh for his Welsh connection :/
FauxFaux
Lucky you.
Muzzz
Or not
my train was JUST cancelled :/
Time to take the long route, sigh
FauxFaux
Ow.
Leaves on the line!
Muzzz
heh
FauxFaux
I asusme you noticed, they have a pair of dates within which they expect leaves on the line.
Iirc it's after 2:30pm on 4th Dec that no more leaves are allowed on the line.
Muzzz
I've seen sheep on the line before
FauxFaux
Welcome to wales.
Muzzz
Only in Wales
Muzzz runs to get the Pendolino to Wolverhampton
stillLotR wonders why Muzzz is going into wales
Girlfriend
I wouldn't bother otherwise , it's shit there
aCiD2
FauxFaux: yessir
stillLotR
ah, so that's how you met her.. the sheep on the line you mentioned before ;)
aCiD2
"Oh, so you're into that TOO!"
Muzzz
Heh, I met her in London not knowing she'd feck off to uni in Wales :|
I feel conned etc
Muzzz also feels FauxFaux's mum
aCiD2
You feel her pain when she sees your face?
Muzzz
"face"
"pain"
inhouseuk
Where the Reading -> London Waterloo line crosses the River Thames at Egham, there are regularly Swans on the line. As they are protected birds, the trains have to stop until they can be tempted to leave.