#metabrainz

ruaok: do you know enough about how ISNI uses our data to answer the latest support email? I don't

mo´´in´

hi!

[critiquebrainz] avma opened pull request #348 (master…master): Add Hebrew Language https://github.com/metabrainz/critiquebrainz/pu...

moooin!

reosarevok: because of the spat between ISNI and MeB, they update manually, thus not very often.

good morning!

ruaok: remind me, what was the problem?

they're being dicks.

self entitled "we are saving the music industry, why are you not bowing down to us" dicks.

Were they the ones who wanted us to pay them to use our data?

yeah.

lol what?

Yeah I don't remember the details but now it does ring a bell (and "lol what?" was my thought at the time as well)

the whole thing started when I found an article in some industry rag that said: "ISNI and YT sign deal. ISNI agrees to get live feed from MB". first i heard of that! that was a bad start.

then they rolled in "we're here to save everything. give us free data and you need to pay $800/year and $ for each registration" and we're a non-profit, so this is all cool.

I looked them up. they are not a non-profit.

Even if they were, why would another nonprofit do that :p

so I offered them a simple deal: We either each pay our respective list prices or no money changes hands. you can choose.

at $18k/year our list is vastly steeper than their fees. so they started wailing.

and then conversation after conversation happend where I felt I like I was talking to a wall.

well an industry wall. "you internet people ruining everything, we shounld't have to pay you"

Didn't you just offer them that they wouldn't have to pay you :D

I did.

now everything I have to deal with them, I dig in further.

if they keep up, every conversation will increase the price/month by $100.

Have you reminded them that "free for both" is still on the table?

The wikipedia page states "it is incorporated under the Companies Act 2006 as a private company limited by guarantee" and then "this UK registered, not-for-profit company"

atj: exactly that. UK company house agrees with that. but they flat out said that they were non-profit.

>10:05:21 AM if they keep up, every conversation will increase the price/month by $100.

😂

I have no patience for assholes.

right then. lets move on. we have loads to do this week.

alastairp: Mr_Monkey: zas: Freso: yvanzo: get your invoices in please. I'll pay the EUR payments tomorrow mornning.

the USD payments are ready to do, so I'll do those and then we can get hacking.

_lucifer: given how late we started in the month, would be it be ok to start paying you in one month's time. the fees for the small amount of the previous month would be pretty high.

ruaok: will do. Alastair and I arrived at the office, see you in a bit

ok.

ok, available invoices paid, packing up, heading out.

shii, how are things over thre, lockdowns and all?

shii, how are things over thre, lockdowns and all?

shii, how are things over thre, lockdowns and all?

shit

sorry

i tried bakcspacing nad hit enter :///

over here it's lockdowns AGIN this week

man am i sick of it

Partial lockdowns here too

(Good afternoon)

ha, good noone here :D

noon

Here they have closed museums and stuff

honestly the library needs ot stay open or i'll go mad

ruaok: yeah, sure

Hmm, I think they did close that here for the month, but not sure

I have way too many books in my backlog as it is :/

alastairp: if you are available, lets work ahead on consul uograde?

iliekcomputers: _lucifer shivam-kapila : we're chatting about the hack week. Should we start a video call so you can join?

yes please.

sure

zoom would work

sure

?

reosarevok: do you have 5 mins to chat about the JS URL issue on beta?

atj: I guess? yvanzo: you around?

resolving the FF44 issue has caused some test failures due to how URL() canonicalises URLs

yup

some of the URLs in the tests are invalid basically

but browsers may be more permissive

Your new tests, or our existing ones?

existing ones

atj: btw, control characters class is [[:cntrl:]] in perl

Examples? :)

hold on

(IIRC the main reason for using RegExp over URL in URLCleanup was browser compatibility.)

e.g. https://www.whosampled.com/Pet-Shop-Boys/Can-Yo...

the URL doesn't work anyway

but ? cannot be used unescaped in a URL

iliekcomputers: _lucifer shivam-kapila : We're ready now, please join https://whereby.com/metabrainz

the working URL is https://www.whosampled.com/Pet-Shop-Boys/Can-Yo...

_lucifer: hi, I'm here, but let's do this planning stuff first, and then look at the upgrade after

atj: when entered in artist’s edit form, we provide this error message: ‘There is an unencoded “?” or “#” character in this URL. Please check whether it is useless and should be removed, or whether it is an error and the URL is misencoded.’

yvanzo: https://github.com/metabrainz/musicbrainz-serve...

so the test is wrong then?

alastairp: sure :)

atj: no, see only_valid_entity_types: []

that means it is not valid for any entity type, so it throws the above mentioned error

yvanzo: btw, https://tickets.metabrainz.org/browse/MBS-9758 is marked as in progress and assigned to you - do you remember why?

MBS-9758: Update Rock.com.ar link format

and the ‘expected_clean_url’ shows that we don’t modify the entered URL so as to let user check it on their own.

reosarevok: yes

yvanzo: OK, thanks that makes sense.

so "new URL(uri).toString()" seems to be encoding those characters

@param your voice is breaking for me. Can you check your microphone once please

iliekcomputers: ^

@Mr_Monkey maybe top recordings for the week?

yvanzo: reosarevok: here are the test failures https://gist.github.com/atj/2363316ac98770057eb...

the spotify failures appear to be a side effect of the escaping of ":"

I'm unsure of how best to resolve this?

atj: about the last example (whosampled with ? in path), why using URL at all here?

I thought that if no change was made, the original URL would be returned.

well, that was the case but it isn't now, however doesn't that just hide the issue?

the URL().toString() is called via cleanURL()

I'm happy to re-add the change detection and return the original URL if no params were removed

Can you please point me to the code? I don’t see it in https://github.com/atj/musicbrainz-server/blob/...

sorry, I haven't merged into the beta branch yet

thanks

what was the issue with this approach?

that approach (which resolved the FF44 issue) causes the test failures

because of the unconditional return of URL.href

I see, it's no longer the same as https://github.com/metabrainz/musicbrainz-serve...

what was the issue with that one then? ^

Basically, the old approach made CatQuest's browser crash every time a URL was pasted

(don't ask about that)

basically, it boils down to the fact that URL.href munges the URL in certain circumstances

the change detection just means we don't trigger that if no tracking params are found in the url

but it doesn't resolve the core issue

sorry, URL.href / URL.toString() are synonymous here

when tracking params is found and deleted, there is no case where it can possibly cause any issue, right?

well, the param deletion is not the issue, the returning of URL.href is.

if I re-added change detection, 99.9% of the time there would be no issue, but if someone used a URL like https://baike.baidu.com/item/啊呀啦嗦?utm_medium=foo they'd get https://baike.baidu.com/item/%E5%95%8A%E5%91%80... back

there is no option to preserve characters encoding?

I mean putting https://baike.baidu.com/item/%E5%95%8A%E5%91%80... in the urlbox or clicking it does go wher it should so

not that I'm aware of. I think the return is correct according to the RFC