bitmap, yvanzo: I noticed one of the top requested file on mb is /static/search_plugins/opensearch/musicbrainz_artist.xml , that's a very small file, which didn't change since a while though we serve it without any client-side caching. What's the purpose of this file? Shouldn't we improve that? Are there any other file of the same kind?
Currently this file only takes 36ms to be served (without any caching) and it fits in one tcp packet, so that's not an issue at all. I just wonder about why we don't have client-caching headers for it and if we should modify nginx config for that.
it was requested 9706 times yesterday (so not any issue either)
bitmap
I have no idea actually
thought it had something to do with FF's search box
zas: I think so, we don't copy it to staticbrainz from what I can see
that'd be a good idea
Nyanko-sensei has quit
D4RK-PH0ENiX joined the channel
Mr_Monkey
Moin moin!
Pac23 joined the channel
pristine__
Moin!
kori joined the channel
kori has quit
kori joined the channel
ruaok
moooin!
Gazooo has quit
Gazooo joined the channel
D4RK-PH0ENiX has quit
alastairp: we're doing catered gumbo on thusday evening the 26th, Nicole is cooking. Care to join us for that?
D4RK-PH0ENiX joined the channel
alastairp
ruaok: sorry, thursday dinner doesn't work for me
ruaok
ok, np.
just sorting out headcounts.
sigh. thpain. "here is confirmation email about your airbnb's. tell us when you're arriving."
I do so.
"no, no, no, contact the owners, not us. the rental agency that isn't the agency with whom you booked."
why the fuck did you even send me mail?
because you wanted to spend more time chatting in email and not actually doing your job?
jwf
cats: RE: GM Last.fm Scrobbler for Bandcamp. Why not Web Scrobbler which supports multiple browsers and sites like Bandcamp? https://web-scrobbler.github.io/
And also scrobbles to ListenBrainz :)
With now-playing notifications even
travis-ci joined the channel
travis-ci
metabrainz/picard#4974 (master - 2bf9891 : Philipp Wolfer): The build passed.
jwf: because "This add-on requires a newer version of Firefox (at least version 53.0). You are using Firefox 44.0."
(great idea though, thnkas for that link)
(I'm working on upgrading, but i'm hittinga bottlenex re tab-groups)
there *was* tab group plugin for 45 but i cant find it anywhere
i ned it so i cna upgrade ot 45 so i nat pgrade to 46 so i cna upgrae to 47 etc.
kepstin
CatQuest: you should be able upgrade a firefox profile straight from 44 to 60esr, then to 68esr without worrying about intermediate versions.
CatQuest
except that it cant becasue 90% of the plugins i literlaly use dayly wont work
i might as well switch to chrome or vivladi (:
kepstin
yeah, hmm. most of them you'd have to find replacements for anyways, which would mean it can't keep the configuration regardless.
CatQuest
also that is too big a change and i cant cope or functino with a browser that big of a change. it's liek movinghouses but it turns out that every soom is upside down and thts a nomral thing but wtf how do i do
yes. but.
if i do it incrementally.
it's managable
sight i get that noone else gets this and thats fien ,but please *try* to udnerstand that i have issues with this
if things change to fast to drastically i cant use it
itry hard ot work aroudn things and b clever but i ha ve to actually use it
i've tried to like, open up latest ff and it's like "..."
ok that was interesting
thne i go bakc and i cna use it
i jsut.. i cna't even
it's lie kthat
i cant even
it's my problem and im broken not the browser (supposedly) but well that's them bones
i don't have adhd and asbergers for fun :(
and. and . the thing is. if i *Do* go trough it, if i do work trough it al land upgrade and reistribute and plugins and all that shit
well thne iu have ot do it ALL OVER AGAIN in 2 weeks
not. worth. it.
:(
kepstin
all the browser and os stuff is kinda tricky nowadays, because either you have to deal with the continuous monthly updates, or you can try to stick to an extended support version and then there's occasional *huge* updates.
well, i guess most of the os's (windows and most linux versions) do twice-yearly updates.
CatQuest
yea
I just want a browser that lets me customise what it looks like, lets me isntall scripts and plugins like adblock and google translate S3 which gets all the security updates and shit But. doesn't. change.the.way.it.looks FFS
instead you can install "themes" which is like fancy as colours and images on the background ಠ_ಠ yea. suuure this is what i mean with "customise hat it loos like" :eyeroll:
sigh
oh and a sane session manager
and tab groups. prefferably
ruaok
ass browser??
samj1912[m]
yvanzo, read the backlog about sir port
I am happy to review the code, but I don't have nearly enough time to port it myself
Might be best to start with tests first and then proceed with the port so we can identify regressions
The port should relatively be straight forward compared to Picard since we don't have to care much about encoding issues
The main problem last time I tried to port it was the rmq part
I think it might be useful to port it to pika instead of amqp
Iirc the new amqp api changed slightly and the old version is py2 only
But I might be wrong
FYI porting it to python 3 might also help with some of the race conditions
jwf
CatQuest: Did you consider a one-time migration to an updated Firefox ESR version? It seems it would better safeguard consistency with a longer supported update lifetime
samj1912[m]
I remember there being some bug in py2 with the interrupt + multiprocessing
I remember adding a lot of flags to deal with this
Which makes the multiprocessing code very complicated
I remember being extremely frustrated with this
The only other source of complication in sir apart from multiprocessing is the dynamic sql queries generated to process updates
I don't think even i remember all the corner cases at this point
But that shouldn't be a blocker for the port
And as for bitmap's suggestion of running multiple instances - that doesn't work with the current architecture of sir. For example if someone updates Bach - processing everything linked to him will take forever for the poor worker that takes the record off the rmq
It should handle normal updates though.
Fixing the above will mean completely changing how sir resolves update links
possibly re queueing everything that's relayed to an entity to rmq again with a different message type instead of trying to process it in a single worker instance
I hope this helps
If you need help on anything else, shoot me an email
yvanzo
thank you samj1912[m]
samj1912[m]
You're very welcome :)
yvanzo
I would still prefer someone else to work on it
samj1912[m]
Just pointers to whoever decides to work on it :)
yvanzo
I can help with by explaining the overall architecture, expected i/o, specs, and other details, but I’m not up to maintain the core.
iliekcomputers
I just read about rms resigning from fsf
Yikes
SothoTalKer
yvanzo: how about writing it up somewhere? (:
yvanzo
"Yikes"?
(to SothoTalKer, about sir)
iliekcomputers: I agree it is a sordid affair :(
SothoTalKer
why yikes? :)
yvanzo
SothoTalKer: "writing [what] up"?
iliekcomputers
Yikes about rms' statements about pedophilia
SothoTalKer
the explanation
yvanzo
iliekcomputers: I mean the whole Epstein affair of course, including this part.
SothoTalKer: oh right, it is partly documented already and I should definitely try to complete it as much as I can.
Plus, samj1912[m] suggested architecture changes, so it is more about discussing architecture changes than explaining what I understood.