(meaning that I think I copied it from the right place)
2013-02-28 05921, 2013
hawke_1
not that I changed it to be what might be more correct. ;-)
2013-02-28 05932, 2013
warp
hawke_1: yes, that is the json i was referring to :)
2013-02-28 05935, 2013
hawke_1
k
2013-02-28 05945, 2013
hawke_1
afk for a few min
2013-02-28 05950, 2013
warp
hawke_1: do you know how to get in this state?
2013-02-28 05951, 2013
hawke_1
warp: Hmm…not really for sure. I … started with 10 tracks from an existing release; added 4 tracks, deleted one from the middle somewhere, added one to the end…started renumbering them with the editable track numbers, remembered that I should be using the arrows…moved some around with the arrows, which is where I think the numbering started getting really weird.
2013-02-28 05927, 2013
hawke_1
…went into the track parser, hit save, found that a track (3) was missing completely, went to the edit note page to see that it was still there…
2013-02-28 05908, 2013
warp
track parser -> parse tracks should also completely re-do the data. odd that that didn't work.
2013-02-28 05917, 2013
hawke_1
Yeah, it still doesn’t work.
2013-02-28 05922, 2013
warp
:(
2013-02-28 05925, 2013
warp
some dodgy code in there.
2013-02-28 05954, 2013
warp
hawke_1: assign the ticket to me, I'll have a look when I can.
2013-02-28 05936, 2013
djce joined the channel
2013-02-28 05906, 2013
hawke_1
If I had to guess I would think it’s something with either the track repositioning not happening correctly when you type it in manually. But that’s just a guess
2013-02-28 05946, 2013
warp
hawke_1: I think a bunch are just suboptimal now that the free-form track numbers are added.
2013-02-28 05914, 2013
warp
(previously you could reorder the track list by editing the track numbers)
2013-02-28 05919, 2013
hawke_1
yep
2013-02-28 05928, 2013
warp
+things
2013-02-28 05944, 2013
hawke_1
That’s what I tried to do, then remembered arrows
2013-02-28 05952, 2013
hawke_1
Does it still even try to do that?
2013-02-28 05904, 2013
hawke_1
(Or maybe does it try to do that even though it probably shouldn’t?)
2013-02-28 05950, 2013
warp
I think I removed it, but I'm not entirely sure. the free-form track numbers were added a while back, and I haven't touched those parts of the code since.
* Ian McEwen: Update the .pot files that don't require migration scripts.
2013-02-28 05959, 2013
MBJenkins
* Ian McEwen: Update instruments.pot partway -- only add new instruments.
2013-02-28 05908, 2013
ruaok
Leftmost: I got the confirmation email.
2013-02-28 05913, 2013
ruaok
where should I forward it to?
2013-02-28 05944, 2013
Leftmost
PMed email.
2013-02-28 05918, 2013
ruaok
forwarded.
2013-02-28 05921, 2013
Leftmost
Thank you.
2013-02-28 05954, 2013
Freso
ianmcorvidae: reo's code wouldn't/won't match "http://ianmcorvidae.net/music/artists/"
2013-02-28 05929, 2013
Leftmost
Freso, but if someone manually selects the BBC relationship, they could enter it that way.
2013-02-28 05944, 2013
ianmcorvidae
yeah, that's not a validation rule
2013-02-28 05949, 2013
ianmcorvidae
that's a match/auto-set rule
2013-02-28 05938, 2013
luks
ianmcorvidae: can geordi do any kind of automatic matching yet?
2013-02-28 05953, 2013
Freso
ianmcorvidae Leftmost: Then we need to update all of the other tests as well.
2013-02-28 05956, 2013
ianmcorvidae
luks: it itself does not, it has an API for people to use to submit automatic matches
2013-02-28 05902, 2013
ianmcorvidae
Freso: most things do not have validation
2013-02-28 05905, 2013
ianmcorvidae
Freso: this is well-known
2013-02-28 05921, 2013
luks
ianmcorvidae: and is the plan to keep it that way?
2013-02-28 05922, 2013
ianmcorvidae
I *just* added it for other database and score, and those have an actual whitelist
2013-02-28 05944, 2013
Freso
ianmcorvidae: I didn't say update all things, I said the other tests (or validations if you wish).
2013-02-28 05946, 2013
ianmcorvidae
luks: well, such scripts could end up in the repository, but as for the geordi web application, yes, the intention is to keep those separate
2013-02-28 05955, 2013
ianmcorvidae
Freso: no, the tests are correct
2013-02-28 05901, 2013
ianmcorvidae
Freso: you're looking at it the wrong direction
2013-02-28 05917, 2013
ianmcorvidae
Freso: how the thing you've quoted works is: if it matches this regex, set to this type
2013-02-28 05930, 2013
Freso
Yes.
2013-02-28 05931, 2013
ianmcorvidae
Freso: while you *thought* it went as "if it matches this type, ensure it matches this regex"
2013-02-28 05948, 2013
luks still needs to figure out what geordi actually is then :)
2013-02-28 05950, 2013
ianmcorvidae
the tests are there to ensure that given a certain URL, it sets the correct type
2013-02-28 05926, 2013
Freso
ianmcorvidae | Freso: while you *thought* it went as "if it matches this type, ensure it matches this regex" --- ... don't tell me what I thought please?
2013-02-28 05928, 2013
ianmcorvidae
luks: it's where the matches are stored, the webapp that lets people submit manual *and* automatic matches, and the importer
2013-02-28 05934, 2013
Freso
Because that was not what I thought.
2013-02-28 05939, 2013
ianmcorvidae
given your comment, that is what you thought :P
2013-02-28 05902, 2013
Freso
Given your statement, I may have explained myself unclearly.
2013-02-28 05919, 2013
Freso
*poorly
2013-02-28 05937, 2013
ianmcorvidae
or misunderstood what reo's patch is doing, yes
2013-02-28 05959, 2013
Freso
...
2013-02-28 05908, 2013
Freso
Okay, so you're referring to the PR now.
2013-02-28 05910, 2013
ianmcorvidae
I don't think that's what you did, but that's the other option :P
2013-02-28 05927, 2013
ianmcorvidae
hence referring to the quote, yes
2013-02-28 05932, 2013
Freso
I was talkign just now, hre on IRC, on changing the other validations.
ianmcorvidae, geordi is trying to use oauth, but it doesn't exist in prod.
2013-02-28 05901, 2013
Freso
In which case I'd say it's out of the scope of reo's patch, as he should follow the code structure of the code already there.
2013-02-28 05902, 2013
ianmcorvidae
ah, so you're suggesting we change the validation rules, which are not tested (by e.g. qunit) at all, hence my confusion, to work more like the current whitelist-style validations (lyrics, coverart, otherdbs, scores)
2013-02-28 05908, 2013
ianmcorvidae
Leftmost: it should be using beta?
2013-02-28 05916, 2013
ianmcorvidae
oh
2013-02-28 05922, 2013
ianmcorvidae
never mind, I see what you mean I think
2013-02-28 05927, 2013
Freso
If we want to change that (which I'm all for!), we should do so in its own ticket.
2013-02-28 05927, 2013
ianmcorvidae
Leftmost: gimme a few minutes
2013-02-28 05932, 2013
Leftmost
Okay.
2013-02-28 05943, 2013
Freso
ianmcorvidae: I'm suggesting that reoafk shouldn't change his patch and that a new ticket should be made for making all validations test the entire URL instead of just the path.
2013-02-28 05944, 2013
ianmcorvidae
I think that ticket he's working on is the same, but specific to BBC music -- I do agree we should do it for other types though
2013-02-28 05939, 2013
ijabz_ joined the channel
2013-02-28 05943, 2013
luks
hm, I think I still don't get the purpose of geordi
2013-02-28 05945, 2013
ianmcorvidae
i.e. since there's already a ticket for "make sure the BBC links are actually the BBC" (the one reo's ostensibly fixing), I'd say we should split on the URL type rather than the type of code change
2013-02-28 05956, 2013
luks
is storing matches actually that useful?
2013-02-28 05921, 2013
ianmcorvidae
luks: well, if you want to use them for something, presumably you need to get them from somewhere, no?
ianmcorvidae: That would still match .*bbc.co.uk/*$
2013-02-28 05919, 2013
Freso
*.*$
2013-02-28 05928, 2013
luks
ianmcorvidae: yes, and that's my misunderstanding. I thought geordi was a tool to import stuff to MB, but it seems a competitor to stuff like the Echo Nest's Rosseta Stone
2013-02-28 05931, 2013
ianmcorvidae
I think it's silly to add code that we know to be not correct just so we can make a different ticket
2013-02-28 05943, 2013
Freso
(Overly simplified regexes FTW \o/)
2013-02-28 05907, 2013
nikki
luks: it's kinda both, isn't it?
2013-02-28 05913, 2013
Freso
ianmcorvidae: I think it's silly to add code that follows a different coding style than its surrounding code.
2013-02-28 05932, 2013
Freso
ianmcorvidae: Esp. when it's for no gain that the rest of the code wouldn't gain from as well.
2013-02-28 05939, 2013
nikki
i.e. I'm keen to get discogs data in so I can find out what's *not* in mb yet, we don't need geordi for the actual discogs matches because those are already in mb itself
2013-02-28 05941, 2013
ianmcorvidae
Freso: too late, our validations are already hugely mutually inconsistent :P
2013-02-28 05943, 2013
nikki
e.g., even
2013-02-28 05907, 2013
ianmcorvidae
there's two perspectives on geordi
2013-02-28 05908, 2013
Freso
ianmcorvidae: That doesn't mean we should make it worse intentionally.
2013-02-28 05926, 2013
ianmcorvidae
one is the editor perspective: I want to import things from these data sources, and mark some things as already done
2013-02-28 05930, 2013
luks
nikki: but if you want to know what's already in MB, you need to do some kind of matching
2013-02-28 05933, 2013
ianmcorvidae
in that perspective, matches are the way to mark something done
2013-02-28 05942, 2013
luks
I don't expect editors to match all of Discogs to MB
2013-02-28 05959, 2013
nikki
luks: yeah, I assumed we'd just automatically do that with a script of some sort
2013-02-28 05902, 2013
ianmcorvidae
the other is the perspective of someone *providing* a dataset, who wants to translate their IDs to MBIDs
2013-02-28 05905, 2013
luks
and I thought the automation of the matching was the primary purpose
2013-02-28 05908, 2013
ianmcorvidae
for them, the matches are the key
2013-02-28 05934, 2013
ianmcorvidae
there is nothing stopping automated matching, either, it's just supposed to be done by external scripts and entered into geordi via an API
2013-02-28 05951, 2013
ianmcorvidae
rather than being done in the request-response cycle of a webapp, which seems like a crazy choice :P
2013-02-28 05955, 2013
warp
the matches are particularly interesting if you have access to a data source other people also have access to.
2013-02-28 05925, 2013
luks
ianmcorvidae: but you can't write such scripts without having access to the data
2013-02-28 05936, 2013
ianmcorvidae
luks: the data that's available via geordi's API? :)
2013-02-28 05903, 2013
luks
ianmcorvidae: so I get get a dump of the geordi index?
2013-02-28 05924, 2013
luks
or how can I iterate over e.g. all albums from the IA?
2013-02-28 05930, 2013
ianmcorvidae
if you'd like, sure -- the search limit result is 10k
2013-02-28 05959, 2013
luks
the point of automatic matching is that I want to match everything
2013-02-28 05903, 2013
luks
I don't know what to search for :)
2013-02-28 05904, 2013
ianmcorvidae
to get a list of IDs, search for * and ask for no fields (and it'll just return IDs)
2013-02-28 05918, 2013
ianmcorvidae
search for everything then! :)
2013-02-28 05939, 2013
ianmcorvidae
a basic script structure might be: search, asking for IDs only, 10k at a time, and iterate through all the pages to get a list of all IDs; then, for each ID, fetch the data, try to match, submit a match if you get one, loop
