I'm actually doing it super-safe: only if *all* names (artist credit, track name of the release, the mediums and the tracklist) are in ASCII, it will set the script
nikki
like 103 releases set to cyrillic which are really just latin
hrglgrmpf
these are errors we only have because we allow setting the script by humans
nikki
yeah, exactly
reosarevok
So, nikki, should hrglgrmpf_bot go wild with setting ASCII to Latin?
hrglgrmpf
I'd say, modbot should set the script automatically for the safe choices
reosarevok
Or is there any possible downside?
hrglgrmpf
reosarevok: it is already going wild :-)
nikki
reosarevok: I already said so
reosarevok
Ok
Just making sure! :)
Yay for botting!
hrglgrmpf
python mechanize is great... I didn't know it before I saw luks script
I had no idea that it is this easy to script the release editor...
I mean to automatically use it
ocharles
mechanize exists for perl too
nikki
:o
ocharles
and 'Network.HTTP.Browser' for Haskell
hrglgrmpf
I've written my own http-handling module once for my b-thesis :-(
ocharles
Doh, I wrote 2 mailing list posts on the train tuesday morning, and forgot to actually send them
hrglgrmpf
maybe we need language guessing, too...
who will categorize 65032 releases without language?
hrglgrmpf: we used to have something in mason that would try and guess the language... it gave some pretty hilarious results
reosarevok
s/hilarious/awesome
Except for actually using them that is
hrglgrmpf
nikki: I'm using pythons guessLanguage for my mails, it works pretty good
I think if all the track names are put together in one long string, it can work pretty well
reosarevok
hrglgrmpf, care to make a test? :)
hrglgrmpf
well, later :-)
reosarevok
Yeah, sure
I mean in genersl
*general even
hrglgrmpf
like, only setting "English" if it is >99% sure would help alot
just to make the report interesting again
nikki
I think something like ocharles's scheduling thing would probably work though, detect the language of a bunch of things, and then show people a screen with the tracklist and the guess where they can say yes or no
reosarevok
Hmmm
reosarevok likes
hrglgrmpf
well, but if I just get 1000 times a normal English tracklist, I will give up because it is boring
reosarevok
hrglgrmpf, add filter-by-language!
People who are bori... I mean, monolingual can use the English list
:p
nikki
well if you detect the language first, you could store the percentage too and make sure people don't get 1000 99% matches :P
unless they want it...
hrglgrmpf
hmm, I always asked myself what "Multiple languages" is giving us
basically nothing...
nikki only uses it for things where there are really a whole bunch of languages
yes, that is good
nikki: hmm, if you feel like it, you can add the license type I think :-)
it has been around for 2 weeks now :-)
nikki
ok
nikki hasn't checked her email today
reosarevok
license!
:)
hrglgrmpf
I'm still not sure how to support it properly by the server though
reosarevok
Yeah
hrglgrmpf
so that download URL and license can be selected both at the same time
but should be possible
nikki
I would -re-add the page we used to have
hrglgrmpf
?
nikki grumbles at myspace and wonders what "Deine Homepage wird routinemäßig gewartet." is supposed to mean
nikki
does that mean they deleted the account?
reosarevok
Oh, hrglgrmpf: I was thinking about the discogs artist edits