WIP: Update dependencies #11
Conversation
I've removed some unused code, name the Kudos system and the statsd integration. A couple libraries are abandoned but still work fine for now. These have been marked in the Pipfile.
LeoVerto
force-pushed
the
update-dependencies
branch
2 times, most recently
from
1586701 to
d630803
Compare
LeoVerto
force-pushed
the
update-dependencies
branch
from
d630803 to
9f105e8
Compare
The previously selected YUICompressor required an entire JVM. This will make the served css 25KB larger but that should be fine for now.
Although for now this ugprade is still blocked by the abandoned django-bootstrap-toolkit.
… migrations The old implementation being used in the migration resulted in data in the admin interface being escaped multiple times. See https://code.djangoproject.com/ticket/27675#comment:8 for a similar problem. Seeing as I had already recreated all the old migrations I decided doing it again would be the cleanest option.
Also found a weirdly similar blog post with nearly the exact same solution.
LeoVerto
force-pushed
the
update-dependencies
branch
from
b3be52e to
a3f7b69
Compare
|
hi @LeoVerto! Sorry, I think I remember when you came in and told us of these updates, but no one followed it up... This PR is a bit of a monster to review, so I think it'd be best if we split it up into a few separate ones (update formatting, update f-strings, update python, update django).... |
|
Yeah, sorry about that. Once I got deep enough into the codebase I just kept fixing things which eventually lead to this monster. I can try splitting it into multiple PRs but prying apart 7facf2d into separate commits for the Python and Django updates would be quite the effort. Would "the big update", f-strings, formatting and docker changes work instead? The reason I never moved this PR beyond the WIP stage was that I never quite got it to work perfectly on my local test setup. The core app and the plugins worked very well but I had some issues with the bot trying to join channels before being authenticated (as described here). brainzbot-bot is built using ancient Go code and uses a homebrew IRC library. Fortunately there is still a Go 1.10 docker container which can run it but ideally most of the code should be thrown out and replaced with libraries. I just haven't gotten around to learning Go yet. :P Other than that BrainzBot should now be mostly up-to-date and somewhat maintainable. You can give it a try by building the docker images from my PRs to the respective repos and starting it using the included docker-compose file. The major missing pieces are documentation and a new CI setup now that Travis is effectively gone. I'd be happy to discuss this further on IRC. |
…tatic Based on LeoVerto's work at #11 thanks Leo! Running in docker means that using whitenoise for static files is easier than serving static files with nginx or other.
I've removed some unused code, namely the Kudos system and the statsd integration.
A couple libraries are abandoned but still work fine for now. These have been marked in the Pipfile.
I've decided to delete and recreate the old migrations. Django was bugging me about adding
on_deleteto theForeignKeyswhich would have made the whole point of automatically generated migrations moot.