#metabrainz

more cats \o/

and reviewed some code, as usual

fin

bitmap: ?

fixed some bugs and perf issues with the json dumps, as found on sentry

also fixed MBS-9342

MBS-9342: InitDb.pl fails to import data with DBD::Pg 3.6.0 https://tickets.metabrainz.org/browse/MBS-9342

and MBS-9309

MBS-9309: Mediums fail to merge when two recording positions are swapped https://tickets.metabrainz.org/browse/MBS-9309

Quesito: https://www.facebook.com/permalink.php?story_fb...

Gracias reosarevok

:)

and MBS-9365 as reported by ferbncode_ (thanks)

MBS-9365: event_meta_fk_id was never created as part of any upgrade script https://tickets.metabrainz.org/browse/MBS-9365

other than that, code review and deployed random stuff

fin, zas

did https://musicbrainz.org/tops/mb_top_stuff.html with the help of bitmap for the jquery part

that top stuff page looks great, btw. thanks!

it's super great :D

!m zas

You're doing good work, zas!

more will come on this

what did you think all my "!m zas" was about? :D

i started to block a bunch of ws abusers (again)

and the spam users are already getting fewer hits. very good

plus usual upgrades and server supervision

fin, samj1912 !

;)

(Still up: yvanzo, psolanki, ListMyCDs, ferbncode; maybe up (please confirm): alastairp, Leo_Verto[m]; anyone else?)

bitmap: np :)

spent time this weekend figuring out mbxml code, apart from that fixed a jira ticket and code reviews

fin

yvanzo: go

salute

I spent most of last week debugging a weird issue in tests renderer and coding the removal of illegitimate accounts.

Also did normal stuff: reviews, forums

finito, ListMyCDs? (psolanki seems to be missing)

I have little concrete to report. I've been toying with the code and it's been challenging but also rewarding. I've learned a lot.

go, ferbncode

Hi :)

his week I mostly worked on the docker image for setting up a MB database with the dump imports. Also worked slightly on dump_manager.py related PR and worked on the final PR to get CB code completely off the ORM: https://github.com/metabrainz/critiquebrainz/pu.... That's all for my week. fin !

anyone left?

alastairp or Leo_Verto[m] maybe

Seems like we're through.

ruaok: spam review

k

yvanzo: what is the general state of the spam stuff?

I'm guessing that bitmap had other things to focus on.

They were *just* here too...

ok, we'll get back to spam related things in a minute.

the PR he has up looks good, I think just tests need fixing

instead, lets go with ws/2.5

It is almost complete, should be in production in two weeks.

ah, ok. back to spam, I guess. :)

how about the script to remove old accounts?

The PR is for half of tickets, I already worked on the rest.

I can hotfix it when it's complete

Spam Spam Sapm Glorius Spam ♫♪♬

bitmap: yes, I'd like that when the other fires are put out.

We can probably push everything this week then.

I'm guessing we won't really have much of a change to review the impact before I leave on vacation on thursday.

does the spammer account removal script have a dry-run option?

♫♪♬ Spam Spam Spam Glorious Spam ♫♪♬

it might be nice to post an excerpt of the output and let people sanity check it.

Freso: If you want to try to discuss multi-level works, ping me.

Sophist-UK: I think it's the only topic for tomorrow, so... :)

yvanzo: please let people sanity check the results before you run it.

+++111

The removal should not be effective before two weeks.

sounds like we're too early to have much of a recap on the spam stuff.

ok, then we can review when I get back.

onward to WS/2.5

zas is having to spend a lot of time blocking asshats and in general combatting spammers.

we really need to move to an API key, there is no way around it MB has become too popular now.

+1

there is a toolkit that can make all of this pretty easy... Kong, right zas?

problem is that it only works with JSON.

yes

I can explain if you want ruaok

sure, go!

Basically Kong is an API gateway, providing everything to manage an API-key based JSON API, including plugins for rate limiting, oauth, and tons of shit

but we can't move to it, because our API is mainly XML

and it would require too much changes etc...

So is it time for a real v3? or why v2.5?

here is the 2.5 trick

we have XML + JSON atm

Fix the two JSON-out-of-beta blockers and make ws/2 JSON + Kong = ws/2.5?

reosarevok: it would actually be ws/3, but I did't want to confuse the topic with the actual ws/3

Exactly !

But doesn't that still kinda force us to keep maintaining 2?

yes.

we're stuck with ws/2 for at least 2+ years.

yes

(But can we kill ws/1 now?)

Freso: different topic, another day.

anyways, that is the general idea.

hi, I'm not really here at the moment, but https://www.3scale.net/ appears to do the same thing from what I understand, and is barcelona local

and we can also move to mbjsonin picard? :P

s/jsonin/json in

alastairp: interesting, we didn't dig that much yet, i guess there are some alternatives

samj1912: yes

samj1912: no reason not to I guess, as soon as the two beta issues are solved?

alastairp: yes, we looked at it many moons ago, but the pricing is shit and traffic has to be routed through them.

kong is much more in the spirit of being independent.

(Also, not FLOSS, FWIW.)

but the idea is that we have API keys and a motivation to move people to JSON.

ruaok: 👍 cool

we can give the ws/2.5 API better rate limits and ws/2 worse rate limits.

so we are switching entirely to a key based api now? or will simply having a key allow more requests?

as a further motivation.

samj1912: just listen!

Younguns are so impatient these days… :)

so, we're floating the idea of this.

okay :P

what do people think?

I think it sounds dandy.

Who will have to get a key - a developer of a tool or every user?

once we have API keys we can do a lot of things.

give people better rate limits and even charge for them.

I just read that as "danky" ಠ_ಠ

Sophist-UK: developers.

zas thinks we can do this without a lot of effort, so we wanted to float the idea and see what people thought.

another thing that we would do is use twitter style rate limiting on the new API.

ruaok: so as a Picard user, I can't pay MB to let me hammer it with my own Picard?

ruaok: Do you envision one key per tool/app, or one key per developer?

charging for rate limits will also include a SLA?

How about a better rate limit for ws/2.5 than for ws/2 to encourage developers to switch?

the same concept we use for LB WS.

I think it's brilliant. and will help in an untold number of ways.

Sophist-UK: " we can give the ws/2.5 API better rate limits and ws/2 worse rate limits." above ;

reosarevok and Freso: Not sure yet. Details to be worked out.

p

Sorry - missed that.

but ideally, it would be nice for a picard user to get a paid higher rate key.

wait.. what now

paying for picard usage??

not that it would bring in a lot of money, but it would shut up some screamers. :)

CatQuest: paying for FASTER usage.

:/

i don't like it

The idea, to start, is to just put a new front end (different url), but keeping the same backend

Picard and Picard Pro. (Same code - only difference is that Pro is paid for and gets a better rate limit.)