logs archiveBotHelp.net / Freenode / #mopidy / 2015 / September / 3 / 1
fu*k, found a bug in dummy audio
which was a root cause for a lot of what I've just been fighting all evening
I don't emit stream changed for the "next" code flow
that greatly reduced the failing tests
basically I think all the remaining ones are due to core not getting audio events without the right hooks
ok guys, i'm on a Arch Linux derivative called Manjaro Linux and I have one problem getting in the way of my music:
[sparky@electrode ~]$ mkdir .ncmpcpp
mkdir: cannot create directory .ncmpcpp: Permission denied
now I fixed that but now ncmpcpp is saying Timeout and got laggy and now i can't listen to music
WARNING Element doesn't implement handling of this stream. Please file a bug.
timeout problem fixed, this is the newest problem ^^
Mopidy-GMusic will be moving to the mopidy org at github \o/
hechtus has basically quit gmusic entirely. new maintainers wanted!
if I had gmusic I would help out :)
I'm planning to get a two month trial subscription to fix gmusic and get a new working release out, then leave it on a back burner
jodal: what I'm considering now is making a "get_core_for_testing(backends=..., config=...)" type helper
basically a sane way for MPD, local, ... tests to get a core instance with correct audio events flow without to much crazy "replay_events" calls in the tests
version I prototyped yesterday before going to bed create a core._test_events, and then core._replay_test_events() gets called on the right code path for making the events just work
so you just need to have dummy audio setup to append events to the core._test_events
hopefully some of this rambling makes some sense
that hack made allmost all the local tests failures dissapear :-)
but first I need more tests for dummy audio as there have been some gaps in the fake due to bad assumptions about always going in and out of stopped
having a day off?
nope, might tomorrow though
mopidy-local-{sqlite,images} plans for the weekend:
- require Mopidy >=1.1 for both
- use Extension.get_data_dir() for images
- drop "Folder browsing" from sqlite, since that just replicates "file" exte
- clean up
- release both as v1.0 to celebrate move to mopidy org
sounds good :-)
jodal: glad that local/data_dir is still there for Mopidy v1.1.1, otherwise the pressure would be too much ;-)
jodal: https://github.com/pyparallel/pyparallel
hmm, I wonder if we can safely get rid of send_async in listener
I suspect the tell() changes jodal did got rid of the need for going via the event loop
mhm, would think so
this would be good because that means testing is easier when actors are running
did a quick test reverting to to the old proxy code and both tests and actual use deadlock
that is with send_async = send
but with the tell code it seems fine
jodal: any objections if I just go ahead an push a change killing that off to develop
better than hidding it in the gapless branch at least
go ahead
with this things basically work out so that you need to have a core with a dummy audio
or else the events to update the state machine in core don't fire
so tests/local/test_playback.py needs to be rewritten to use a core actor, not core directly
or use one of my hacks
mhm, I guess that's ok. it was just a minor performance hack to not use a full actor there.
« prev next »