I mostly did support, editing, and trying to enjoy the last days of summer (e.g. more camping)
I also helped CatQuest a bit (with a delay)
Dr-Flay
You can't tell if an experimental musical instrument will become more than an oddity, until it is no longer experimental and more people use it. eg the Theramin
reosarevok
Dr-Flay: hence "demonstrate actual usage" :)
Dr-Flay
by their nature an experimental instument may be unique
reosarevok
That's about it for now, but summer is finishing sadly so I'll be back to staying at home looking at the rain
slash screen
So you'll see me more, probably!
Freso: you? :)
Dr-Flay
well, youtube has some weird experiments, like the dude that made a guitar from old 8bit kit
Freso
See you more, or see more you? ( ͡° ͜ʖ ͡°)
Anyway.
reosarevok
Dr-Flay: agreed, but let's continue after the meeting (so, in like 30 min maybe? :) )
Freso
🙋
Yes plz.
So I herded and reviewed a number of final GSoC blog posts.
I haven't been much involved with GSoC otherwise, so it was great to get this chance to get a slightly more in-depth look of what y'all've been up to the last few months. Great stuff. :)
I also registered myself as attending the GSoC mentor summit as our designated GCI mentor/admin, so after seeing iliekcomputers and yvanzo in Barcelona in 1½ months, I'll be seeing them again in ~2 months. :p
iliekcomputers
Freso: <3
looking forward to that :))
yvanzo
See you soon!
Freso
Other than that, mostly being around and about, handling spam, etc.
fin; yvanzo, go!
yvanzo
Hi!
Released a MusicBrainz server update along with a new slave server virtual machine.
JaredCE has quit
Freso
(Only bitmap and zas left on my list. So few people now all the GSoC'ers are no longer obligatorily giving reviews. :'( Remember that you're free to give reviews regardless!)
yvanzo
Reviewed GSoC project SpamBrainz by Leo__Verto, and pull requests by bitmap.
Worked towards JIRA SSO, boosted by legoktm's own work towards Wiki SSO. :^)
Freso
yvanzo: NicE!!!
yvanzo
Continued converting UI code to React, and updated guess case/guess featured artists.
Freso
*e
!m yvanzo & legoktm
BrainzBot
You're doing good work, yvanzo & legoktm!
yvanzo
Acabado, go bitmap?
bitmap
hey
kartikeyaSh
Freso: i'm up
bitmap
mostly worked on some prs for the react migration, submitted one for the artist merge ui and converted some knockout bits
fixed some bugs related to the recording video icons and standardized where it's displayed
Leo__Verto is now known as Leo_Verto
looked into some memory spikes with the webservie containers and found an issue with the /ws/2/discid endpoint, submitted a pr to limit the amount of data it returns
zas
\o/
bitmap
just code review other than that. fin, zas?
zas
Hey
we had issues with burnside, which was replaced by a better server, most of the hassle handled by ruaok & hetzner techs
rsh7
Freso: i'm up for the review.
zas
also we had issues with diskspace on prince, due to some obscure bug in docker, fixed now
renewed *.mb.o cert and deployed
migrated all domains to gandi V5 (and what they call LiveDNS)
Picard PRs reviews & patches
merged and deployed picard plugins fixes
plus usual stuff, upgrades, and a lot of MB edits
fin. kartikeyaSh ?
kartikeyaSh
hi
Freso
(Only rsh7 left on my list now. Last chance for anyone else to speak up!)
kartikeyaSh
This week I didn't do anything on MessyBrainz. But I read a lot of data structure problems for the interviews. Now I can create a linked list from a binary tree (without using extra space). I can convert stack into a queue blah blah. Don't know why on earth I'll do that. Once I'm free I'll work on MsB PRs to get those merged.
ruaok also prods LordSputnik and yvanzo about gsoc evals
fin
bukwurm
Freso: Me!
kartikeyaSh
rsh7: go
rsh7
Hey!
LordSputnik
ruaok: I'm writing my feedback now :)
rsh7
Last week I filled the final evaluation and completed the blog post.
And then came back to work on open PRs (before gsoc). I would complete some left over work of AB-98 this week now.
And regarding the evaluation PR, after I get a review from iliekcomputers , we'll complete it soon.
fin.
bukwurm: ?
Freso
bukwurm: Go. :)
iliekcomputers
rsh7: asap, sorry. I totally forgot about that with the dump PR and the spark stuff.
bukwurm
Hey everyone!
This week I worked on the PR reviews.
Fixed some bugs
Working on some remaining features
rsh7
iliekcomputers: no issues 😅. Let's get it done this week.
bukwurm
Most of the PRs are merged now
Some changes are remaining on the site, part.
Freso wonders if Leo_Verto / Leo_Verto[m] wants to chime in with something too
Which I aim to finish off this week.
That's it for me.
Freso
Alright.
LordSputnik
bukwurm: did you see my comments about the memory leak? I managed to get stuff into the queue now, but the consumers are encountering lots of errors to do with the link_import table PK being violated
bukwurm
Yeah
Freso
No one else has said they want to give a review and we don't have any further items on the meeting agenda for tonight.
LordSputnik
I can go
Freso
So thanks for your reviews and for your time... alright, LordSputnik, go. :p
LordSputnik
Haha
This week I worked on entity editing
I managed to successfully add relationships between entities on my local test site
So the new relationship editor stuff is finally working
I then worked out how to get it to work properly for adding relationships at the same time as entities are created (rather than just for existing entities), so I'm working on doing that now
As a result of the improvements to editing I've been doing, we can now cut about 200-300 lines of code from the entity editing logic
Which is a nice side-effect :)
Freso
🙌
LordSputnik
Apart from that, I've been reviewing bukwurm's work and also talking to Mr_Monkey about his work on webpack
iliekcomputers
rsh7: 🤟🏽
LordSputnik
fin.
Freso
Okay.
No people spoke up during LordSputnik's review.
So I think we'll actually really be ending here. ;)
Am I free to suggest some experimental electronic instruments if wanted ?
reosarevok
Dr-Flay: rather on #musicbrainz, but probably? :)
Dr-Flay
sorry just dealing with a masterserver support case
Leo_Verto
Oh hey, I missed the meeting! :D
anyway, not much from me last week (spent most of it working on home automation) but it looks like there's already someone interested on working on SpamBrainz for GSoC 2019 :D https://blog.musicbrainz.org/2018/08/14/gsoc-20...
Gazooo has quit
Gazooo joined the channel
bukwurm
LordSputnik: Ok, so primary reason behind the errors is perhaps due to the reimporting of the same record
i.e. the record still exists
So, to handle this - we can perhaps add a check to get the link import record before pushing into the db
That way we can also ensure the timestamp is the latest
The error probably should go away once the primary key is unique for the records we are importing
I've updated the config sample
For handling the memory leak, we need to have a discussion on how to approach it in the best possible manner
yvanzo
Leo_Verto: It looks like the blog post URL changed since you submitted it to GSoC, it used to be on 2018/08/13, right?
bukwurm
I had a small plan, which I mentioned couple of days back
It was to use confirmChannel instead of simple channel
in rmq
It basically returns promise which resolves when the data has finally reached the queue
In this way, we can either pause the stream while a single message or a bulk of specified size has reached the queue
And then resume it later
LordSputnik: If there's any other way to handle the primary key errors on the consumer side in your mind, let's have a discussion any time your free.
*you're
LordSputnik
bukwurm: yes, I think that's what has happened - it's violating the unique constraint, not the "not null" constraint
as for the leak - having promises which resolve when the data reaches the queue sounds like a good start to a solution
Dr-Flay has quit
I'm not sure though
bukwurm
The reason why pausing the stream when the buffer is full will not work is, by default the size of the stream is 1.5 G per process. It will wait until all available buffer upto 1.5 is filled before returning false for buffer availability.
LordSputnik
Ideally we'd have a pool of outstanding promises and then only read new lines once promises started being resolved
The size of the "stream" that the library is mimicking for sendToQueue?
bukwurm
LordSputnik: Yeah, that's what I meant by awaiting for a bulk of messages.
The default nodejs stream size
I think they're using standard streams, no?
LordSputnik
Going back a step - are we 100% sure what the cause of the problem is?
bukwurm
Anyhow, it didn't work when I tried to pause upon sendToQueue returning false.
LordSputnik: I am not 100% sure, but I definitely think that's the problem.
LordSputnik
Is it easy to modify the producer to run in the main process? Then we can profile the memory usage much more easily, I think
bukwurm
Disk I/O using streams exceeds Queue I/O
Modify how?
LordSputnik
Or have a separate testing entry point which just calls the producer with some arguments from the command line in a single thread
Leo_Verto
yvanzo, ah damn, I forgot the URL included the date. can we get that changed somehow?
LordSputnik
Then we can run that with node --inspect, and view the memory allocation in Chrome Dev tools