samphippen did all sorts of unspeakable things to that dialog. feel free to restore sanity.
2017-03-12 07135, 2017
Zialus has quit
2017-03-12 07139, 2017
Zialus joined the channel
2017-03-12 07150, 2017
github joined the channel
2017-03-12 07150, 2017
github
[listenbrainz-server] paramsingh opened pull request #142: LB-127: Make position of last.fm importer dialog box fixed (production-docker-setup...importer/box-position) https://git.io/vyKlx
ruaok: I'll send a PR in a few hours, right now have to go afk for some time.
2017-03-12 07151, 2017
ruaok
LB currently uses redis to build a data flow pipeline that sends data to the influx database, where duplicates get removed.
2017-03-12 07120, 2017
ruaok
using another pipeline, the deduped listens get sent to bigquery.
2017-03-12 07150, 2017
ruaok
if we stop the redis container or it crashes while people are streaming listens, we would lose data if we have no persistence.
2017-03-12 07152, 2017
ruaok
it seems that turning on AOF persistence is the right thing in our case -- assuming the docker volume is preserved, re-starting the redis container should re-store the right data values and allow the pipelines to continue.
2017-03-12 07109, 2017
ruaok
thoughts? bitmap, zas, Gentlecat, kepstin, kuno?
2017-03-12 07156, 2017
Lotheric has quit
2017-03-12 07105, 2017
iliekcomputers thinks taking snapshots every X time intervals along w/ aof wouldn't harm either
2017-03-12 07108, 2017
iliekcomputers
but I am no expert
2017-03-12 07149, 2017
ruaok
would slow things down. we don't really care about any point in time -- that data will not be useful later on.
2017-03-12 07156, 2017
ruaok
we only care about persistence between restarts.
2017-03-12 07106, 2017
samj1912_riot[m]
zas ping
2017-03-12 07118, 2017
samj1912
I had a discussion about 3.6 and 3.4 with jrobenson and wanted to discuss out target version
2017-03-12 07116, 2017
zas
samj1912, ruaok : pong
2017-03-12 07121, 2017
ruaok
aof_enabled:1 :)
2017-03-12 07138, 2017
ruaok
zas: I need to copy of json file from docker-server-configs to consul.
2017-03-12 07103, 2017
ruaok
I do NOT want the json file interpreted -- it needs to arrive as a string (really, a json file) inside the container.
2017-03-12 07110, 2017
ruaok
this is for the bigquery-credentials.
2017-03-12 07111, 2017
ruaok
effectively copy docker-server-configs/consul/LB/bigquery-credentials.json as /code/listenbrainz/credentials/bigquery-credentials.json
2017-03-12 07115, 2017
ruaok
any idea how to do that?
2017-03-12 07115, 2017
zas
hmmm, store the json data in a json file as it was a "normal" value ? Dunno if there is any another easy mean to do it ?
2017-03-12 07140, 2017
ruaok
encode into json and store in json?
2017-03-12 07145, 2017
zas
yes
2017-03-12 07151, 2017
ruaok shudders.
2017-03-12 07151, 2017
zas
as a normal string
2017-03-12 07104, 2017
ruaok
ok, then on the flipside...
2017-03-12 07110, 2017
zas
dunno how to else, and not even sure it works
2017-03-12 07140, 2017
ruaok
I wonder if that file will arrive decoded.
2017-03-12 07113, 2017
zas
can you just parse your json file as usual with git2consul and use this data to build a json file from a template in your container ?
2017-03-12 07112, 2017
samj1912
zas: so basically, python3.6 pros are that it handles the windows encoding issues related to gh:p#605,607
zas: yeah it is very dependent on the structure of the json file. it shouldn't need to know. but it should also work. :)
2017-03-12 07106, 2017
ruaok
let me try it.
2017-03-12 07119, 2017
Mineo
samj1912: re antlarr's pull request: I can't start picard on that branch at all because an import of get_file_path which is not defined in picard/util/__init__.py
2017-03-12 07101, 2017
samj1912
for windows binaries, it will be better to have python3.6 included in them, the only cons are that *nix wont support 3.6
2017-03-12 07113, 2017
samj1912
but since we are talking about a year in future...
2017-03-12 07131, 2017
samj1912
antlarr: ^^ rebase issues, please fix
2017-03-12 07150, 2017
Mineo
also, could someone tell me what the current merge-via-github and branch settings for picard are?
2017-03-12 07141, 2017
zas
samj1912: did you evaluate the work needed to be compatible with 3.4 and 3.6 ? perhaps it will be simple enough to provide alternative code depending on versions without too much hassle
2017-03-12 07106, 2017
zas
of course, if needed changes are too important, we'll have to choose
2017-03-12 07111, 2017
malhotrag joined the channel
2017-03-12 07122, 2017
samj1912
zas: basically all the sys encoding hacks should go since they are implemented in python
2017-03-12 07147, 2017
samj1912
afaict from my discussion with jrobenson
2017-03-12 07148, 2017
zas
the main issue is the network side, where 3.6 has some unique features
2017-03-12 07109, 2017
samj1912
I'll go through the changelog for 3.6 once
2017-03-12 07122, 2017
samj1912
and add the evaluations to my proposal
2017-03-12 07131, 2017
zas
ok great
2017-03-12 07101, 2017
samj1912
also when you are free can you look at the one on discourse?
2017-03-12 07104, 2017
samj1912
I updated it
2017-03-12 07121, 2017
Mineo
note that python 3.4 is not even in bugfix mode anymore, it's security fixes only
2017-03-12 07100, 2017
samj1912
yes
2017-03-12 07110, 2017
SothoTalker_
you could always use the codecs module to handle stuff like unicode in- and output
2017-03-12 07134, 2017
samj1912
SothoTalker_: yes, that's a hack
2017-03-12 07150, 2017
SothoTalker_
why?
2017-03-12 07101, 2017
samj1912
because it still causes issues
2017-03-12 07116, 2017
samj1912
I believe I linked the issues the last time we had a discussion on that PR
2017-03-12 07142, 2017
SothoTalker_
3.6 can be used safely since the distributed program will be independent from any other installed python version :p
2017-03-12 07131, 2017
samj1912
for the OSes we built binaries for, yes. linux?
2017-03-12 07150, 2017
SothoTalker_
we can build a binary for linux, too
2017-03-12 07124, 2017
Mineo
let's not go down this route, please
2017-03-12 07116, 2017
samj1912
I dont want to either
2017-03-12 07125, 2017
SothoTalker_
scared of executables? :x
2017-03-12 07131, 2017
samj1912
as someone who has been a part of debugging all the packaging issues we have had with windows and OSX in the last 4 months, I dont want another headache..
2017-03-12 07151, 2017
Zastai joined the channel
2017-03-12 07111, 2017
SothoTalker_
less alcohol while debugging ;-)
2017-03-12 07107, 2017
Mineo
are you volunteering?
2017-03-12 07110, 2017
samj1912
I dont mind it if you offer to package picard for all *nix distors and take care of all the issues that arise :P
2017-03-12 07140, 2017
JonnyJD joined the channel
2017-03-12 07150, 2017
SothoTalker_
why should I care about linux :D
2017-03-12 07108, 2017
SothoTalker_
the I is capital and bold and underlined
2017-03-12 07140, 2017
SothoTalker_
make a list what linux distro uses which python by default and then decide if it's worth it or not ;)
