Another email by someone thinking the "authority control" link to MB on a Wikipedia page means we run the page and they need to ask permission from the "authorities" to change stuff :p
2017-07-26 20703, 2017
zas
samj1912: i removed the required attribute for appveyor "branch" (and kept "pr")
2017-07-26 20742, 2017
zas
ruaok, reosarevok: it seems to confirm we don't lack stupid people on Earth ;)
2017-07-26 20722, 2017
ruaok
there is only one thing more abundant than hydrogen in the universe: stupidity.
2017-07-26 20701, 2017
psolanki
Gentlecat: rendered the rating display and necessary changes in form
2017-07-26 20702, 2017
psolanki
Styling part is to be done
2017-07-26 20703, 2017
psolanki
will make a PR soon
2017-07-26 20737, 2017
Gentlecat
ok, I'll be waiting
2017-07-26 20751, 2017
UmkaDK has quit
2017-07-26 20752, 2017
arbenina_ joined the channel
2017-07-26 20753, 2017
ruaok
zas: we currently have two musicbrainz.org SSL certs. mb.org and *.mb.org. mb.org is up for renewal soon. I believe we're not using that one. confirm?
2017-07-26 20722, 2017
UmkaDK joined the channel
2017-07-26 20745, 2017
zas
let me check
2017-07-26 20756, 2017
zas
confirmed, we don't use mb.org one, only the wildcard one
2017-07-26 20709, 2017
ruaok
oki
2017-07-26 20714, 2017
ruaok
9,845,364!
2017-07-26 20716, 2017
ruaok
so close.
2017-07-26 20717, 2017
MajorLurker has quit
2017-07-26 20717, 2017
iliekcomputers
:)
2017-07-26 20757, 2017
drsaunders joined the channel
2017-07-26 20702, 2017
agentsim joined the channel
2017-07-26 20750, 2017
samj1912
zas: so, I have completed most of the network related refactoring required, I was wondering what direction to take to improve UI responsiveness and I/O. The main problem is that the tips in Qt dont exactly apply to pyqt because it is not able to multithread in essence because of GIL in cpython.
well, I think lazka said something about mutagen not occupying much cpu time, but that doesn't correspond with what we see in picard
2017-07-26 20753, 2017
samj1912
the ui becomes generally unresponsive while loading files
2017-07-26 20758, 2017
samj1912
and while saving a ton of them
2017-07-26 20737, 2017
samj1912
this might be because of how we have implemented the threading code and the fact that the UI thread runs on the same thread as loading thread
2017-07-26 20748, 2017
samj1912
however that does not explain the slow down on saving
2017-07-26 20749, 2017
github joined the channel
2017-07-26 20750, 2017
github
[listenbrainz-server] mayhem closed pull request #236: Make the API blueprint use a url prefix when registering (master...api/prefix-var) https://git.io/v7fxi
2017-07-26 20750, 2017
github has left the channel
2017-07-26 20719, 2017
ruaok
alastairp: your last import stamp says you haven't imported into listenbrainz yet. is that the case or is that a bug?
2017-07-26 20703, 2017
samj1912
zas: ?
2017-07-26 20728, 2017
bitmap
samj1912: didn't Sophist-UK do a lot of investigation into that? I don't really remember how it's implemented anymore
it was very buggy when it came to handling signals
2017-07-26 20755, 2017
samj1912
picard wont close on ctrl+c anymore
2017-07-26 20702, 2017
Sophist-UK
But it was one of those problems where as you dig you find more issues and then it gets messy and in the end times out.
2017-07-26 20708, 2017
samj1912
yes
2017-07-26 20726, 2017
samj1912
and I am not sure about the random processevent calls we had
2017-07-26 20754, 2017
samj1912
and the priority queue you had implemented
2017-07-26 20754, 2017
alastairp
ruaok: imported
2017-07-26 20700, 2017
alastairp
exactly the same number of listens
2017-07-26 20704, 2017
ruaok
great.
2017-07-26 20715, 2017
alastairp
apparently none of my music players are scrobbling to lastfm any more
2017-07-26 20719, 2017
alastairp
but that's not your fault
2017-07-26 20723, 2017
Sophist-UK
1. ProcessEvent does not work the way thread.py thinks it will.
2017-07-26 20741, 2017
ruaok
I'm itching to make spotify recording work with the new endpoint hugh hacked up.
2017-07-26 20742, 2017
Sophist-UK
You can only call ProcessEvent when you are on the main thread.
2017-07-26 20758, 2017
ruaok
not sure how well it can scale or how auth works, but I am dying to dive in.
2017-07-26 20704, 2017
Sophist-UK
What you need to do is a micro-sleep on the worker thread to give the main thread a chance of getting a look in.
2017-07-26 20727, 2017
samj1912
Sophist-UK: there should be a better solution for this than introducing sleepers to pre-empt threads
2017-07-26 20754, 2017
Sophist-UK
On Py 3.2+ (?) - anyway for Py3 we are using - GIL is better at choosing which thread to run and in particular better if you set O/S thread priorities.
2017-07-26 20708, 2017
Sophist-UK
samj1912: Not as far as I am aware.
2017-07-26 20717, 2017
chirlu has quit
2017-07-26 20735, 2017
Sophist-UK
You can adjust how frequently it interrupts threads to allow another one a chance to execute.
2017-07-26 20751, 2017
Sophist-UK
But Py3 is better.
2017-07-26 20714, 2017
zas
Py3.5+ will be used
2017-07-26 20721, 2017
zas
that's for Picard2
2017-07-26 20758, 2017
samj1912
Sophist-UK: there are also those recursion errors that happened when you tried all of this
2017-07-26 20714, 2017
Sophist-UK
In #604 I did a micro-sleep just before a worker thread terminated (which obviously has little impact) in order to prevent the next worker thread from executing instead of the main thread.
2017-07-26 20719, 2017
Sophist-UK
Back in a few mins - customer here.
2017-07-26 20748, 2017
iliekcomputers
ruaok: what was it you wanted to talk about?
2017-07-26 20719, 2017
ruaok
making the switch from beta to production.
2017-07-26 20740, 2017
ruaok
got a minute or two to work on a switchover plan?
2017-07-26 20750, 2017
iliekcomputers
Yeah sure
2017-07-26 20751, 2017
chirlu joined the channel
2017-07-26 20735, 2017
samj1912
zas: just fyi, I worked on gnome music for a while and they use a tracker to load the mp3 files in bg, we can look into IPC instead of threads
2017-07-26 20754, 2017
samj1912
but that code might be a bit out of scope for this gsoc project
