#musicbrainz

KyleB (~Kyle@dhcp065-024-013-066.columbus.rr.com) has joined #musicbrainz

icepick-away (ytilezejow@icepick.info) has joined #musicbrainz

i just saw a comet

!

or a start falling

JamesConf (~james@rdu74-173-140.nc.rr.com) has joined #musicbrainz

anyone mind helping me with a little config proble with trm?

aeon (aeon@c-67-164-29-92.client.comcast.net) has joined #musicbrainz

aeon_t (aeon@c-67-164-29-92.client.comcast.net) has joined #musicbrainz

hey all

anyone know what "Failed dependency" error means on my failed moderations?

djce (~davide@195.60.9.122) has joined #musicbrainz

Hi aeon_t. Still there?

Well anyway here's the answer to your question, you can read it when you get back.

"Failed dependency" means that something to do with the data you were trying to change has itself changed since you entered the moderation.

For example, let's say you change an artist name from "Bob" to "Bobby"

If someone else changes the name to, say, "Robert" before your mod gets approved,

then you'll get a Failed Dependency.

This is because the system checks that the old data is still in place ("Bob") before changing it to its new value ("Bobby").

But in the example, it's not "Bob", it's "Robert". Hence the failure.

Hope that helps!

If you could tell us exactly which moderation you're interested in, we could probably give you a more precise reason.

ah, I see.

thanks a lot.

aeon_t is now known as aeon_sleep

alus (alus@cs6668110-201.austin.rr.com) has joined #musicbrainz

djce (~davide@195.60.9.122) has joined #musicbrainz

ruaok (~robert@adsl-63-194-101-14.dsl.snlo01.pacbell.net) has joined #musicbrainz

djce: evening!

hi!

did you pick up the changes I checked in last night for libtp?

Not yet.... let me go see.

not important, I just wanted you to be aware.

ah yes, new calls. ok.

how's the TP-tagger doing?

forming into shape?

Coming along.

Its not too much work, really.

All the hard stuff is being done by libtp.

Good :-)

Now its just UI, which is nice.

wxPython, yes?

oh, no.

oh no, C first isn't it.

I remember.

Yes, its mb_tagger rehashed.

I won't start on the python tagger until after BM.

ok.

I can imagine it feels pretty good to go through the mb_tagger and delete large swathes of now-ununsed code.

me too.

There is less satisfaction in deleting whole files.

I like both.

That was the biggest case -- the code base is only 1/3 of the previous...

:-)

What is tp_SimilarityCheck? Is it a "do" method, or a set, or get, or what? I can't tell from the code.

its an 'or what'.

:-)

oh, that's all right then :-)

Given a threshold it checks to see if there are any matched files that are below that threshold.

for all tracks?

in order to pop up a warning for low matching files.

All tracks that are recognized

and that's the return value? if the popup should be shown?

t/f

yes

yeah, but I can't think of any!

tp_AreAnyFilesBelowSimilarityThreshold :-(

int tp_CountFilesBelowSimilarity ?

Anyway, I've just committed the perl binding changes for the two new methods.

yeay!

ok, time to hit the road...

Talk to you.... probably sunday or monday...

l8!

codeFiend (aeon@c-67-164-29-92.client.comcast.net) has joined #musicbrainz

hmm, does anyone know what the internationalization plans for the database are?

What sort of things do you have in mind?

well, for example artist names in non-latin charsets.

That's already possible.

I tried searching for several russian artists in cyrillic - no luck.

Certain parts of the system are either optimised for, or at least work best with, latin / English data though.

hmm.

Are you sure it's in there?

I mean, is it the search that's faulty?

no, I'm not sure. But if search can't turn it up, I can't verify that it exists.

True.

and since the 'browse by alphabet' function is for latin alphabet only, I can't use that either.

Currently the search works like this:

your search text is split into words, on non alphanumeric characters.

(so that should still work in Cyrillic).

I could try creating an artist in cyrillic and then doing a search for it... and seeing if that would turn it up.

Then, the words are converted to lower case.

which should also still work.

Then, it is "unaccented" - which might be the iffy part.

Then it's matched against the artist names.

hmm.

The "browse" pages are one of the few parts of the system that only really work with Latin stuff.

right.

*ponder*

well, let me try and do a test run with adding a cyrillic artist and then searching for it...

ok

just a minute.

I'd be interested to see how it goes...

I know there's non-latin stuff in there, but I'm not sure what difficulties are likely to be encountered.

hmm.

well, adding it worked.

let's see if it turns up in search

what's its URL?

woot. it actually did.

color me surprised ;)

Cool !

UTF-8 rocks ;)

I'm fairly sure there are some texts where the "unaccent" thing is decidedly questionable.

hmm.

Some scandinavian stuff, I think.

But the whole idea of "unaccenting" sounds dodgy to me. It's inherently not locale-safe.

yeah.

definitely.

ok, well, now I'm wondering.

But I'm glad the search works.

hmm.

gimme a minute here...

hmm... ok.

now I'm seeing a different problem with this, conceptually

I just added russian artist Aquarium, (in cyrillic) and put in a mod to merge the current listing into it... now, if an english-speaking person wants to find it, but they don't know how to spell it in cyrillic, are they going to be pretty much farked? perhaps a 'transliterated title' field should be added to the db?

or perhaps just adding an alias for the artist with the transliteration would be enough?

It's an idea.

aliases would work too.

hmm.

In fact aliases are probably the better way to do it.

more generic, true.

hmm.

that works for artist names

:-)