#musicbrainz

alus_ is now known as alus

hello from Argentina

hi!

greetings from California!

I know the place, nice....

I've never been to .ar :-(

we have a lot of places quite "Californian"...

that's why we have such god wines :D

good I mean...Sorry God

heh. I prefer .ar wines over .ca wines any day.

yes Gallo it's just begining... 9.9

As in E&J Gallo? That's not wine.

I live smack in the middle of wine country.

We practically have it coming out of our asses. :-)

well, Iwas trying to be friendly... I didn't know you were a wine expert

I wouldn't call myself and expert, but around here its hard to avoid. :-)

I know it's like the people here living in Mendoza or San Juan states (provinces here)

I was exploring rooms, testing Colloquy...

Knio is now known as Knio-sleep

alus_ is now known as alus

yawn

run naked through the streets

lol

too cold :-)

we just went for our inaugural soak in our new hot tub.

what an ordeal getting that installed, but its finally done.

not pleasant either

nope.

painful, indeed.

I'm glad someone explained me the point in XML

very embarrassing that I didn't get it after all those years ;->

eh?

which point?

that you can add elements later and don't break existing parsers :-)

oh.

ya. :-)

I learned SGML 4-5 lives ago.

Horribly complex, but pretty cool.

xml is all the best of SGML.

much simpler

so when xml came around, I was sold, right off the bat.

pure horror

I never even heard of that. What timeframe was that?

it was quite popular 7-10 years ago

it's a lisp dialect

ick.

people tried to use it for XML, too

with emphasis on 'tried'

I was never fond of lisp, ml or prolog.

lisp is kind of cool for toy problems

I used to have Turbo Prolog, but that was the ONLY language I never managed to write a meaningful program in.

anyway, I think I'll hack on mb_index today

:-)

in the evening I think I'll have updating via XML packages implemented

I was just thinking the same. I want to finish the last set of codings for libtp.

testing takes much, much longer

Go to bed, debug it in the morning.

:-9

:-)

sounds like we're in similar places.

I can't wait to go back to python.

And my first goal is to replicate the tag editing features in easy tag.

Just have a solid tagging app, and then load the cool identification features on top.

apt-get install easytag

whoo. could it be so easy :-)

its not the greatest program, but it conveys some important features.

:-)

ok, I'l off to coding.

ok, have fun!

s/l/m/g

du auch!

danke

yalaforge: you still around?

yup

can I bounce an idea off you?

sure

ok, think libtunpimp.

pipeline architecture.

ok

we used to have an analyzer that did metadata read and TRM at the same time.

since picard will do more album oriented lookups, the need for TRM is less.

Thus metdata read and TRM have been decoupled.

so you want to split the thread, too?

yes.

sounds sensible

keeping functional blocks seperate

that's all pretty much past tense.

the question I have is this...

It used to be that a track through libtp had a certain path, and all the transitions were obvious.

pending -> trm lookup -> (rec, unrecog, error)

but now we have:

readmetdata -> unidentified or TRM

but the read metadata doesn't yield any info to make this decision.

The caller needs to direct the flow of the track.

Is this track TRM bound?

Or just readmetadata -> recognized or unrecognized?

does the problem make sense?

I'd readmeatadata, then query the web service

if unrecognized, then TRM

well...

kinda.

that way it stays a linear pipeline

you have to add a number of tracks and then perform that album lookup on those number of tracks.

so where do some tracks go while new tracks are being added?

you need a thirds state (recognized, unrec., accumulating)

right?

now I am wondering if I make too mcuh of this.

seems like its getting complicated.

hm ...

The read metadata should move a file to either recognized or unrecognized.

The caller should watch the messages for tracks that appear unrecognized.

And move them along by doing:

1. nothing.

2. clustering into an album and doing and album lookup while it stays in unrecognized.

3. set to pending for TRM generation.

Yes, that makes sense.

The tagger will need to get modified to look for unidentified files that have no TRM.

For those files, just move them to pending and keep chugging.

:-)

Thanks for listening. :-)

np :-)

lol