Create dramatic, zoom in gifs! This is what runs http://dun-dun-duh.com/.
- Python ~2.7
- gifsicle
- Redis
DDD does not have very many configuration options. These can be placed in a python file with a path specified by a "CONFIG" environment variable.
This is the timezone you want your statistics recorded in. Since stats are driven by dates, you'll want to set this to your local timezone.
DDD expects the system to be configured to UTC time. If you don't, your statistics will be weird.
This is a Flask config variable, but necessary for DDD because we use sessions.
This is a connection string in a manner that flask-and-redis can grok.
Defaults to redis://localhost:6379/0
This determines whether completed gifs are stored on the local system, or S3.
Valid values are "s3" and "local". Defaults to 'local'
These are your AWS key and bucket name if you are storing gifs on S3.
This is a brutal way to "fix" URL's for gifs. It allows you to set a format string for a URL to a gif.
This value is "http://i.dun-dun-duh.com/%(filename)s" in production, because we have i.dun-dun-duh.com CNAME'd to the S3 bucket domain.
This is the path where we will store temporary upload files, and, if UPLOAD_DESTINATION is "local", output gifs.
Defaults to a folder named "upload" in the same directory as dundunduh/__init__.py
Assets are in a half-broken state. Sorry.
Right now, you need to rebuild them with Grunt when you change them.
DDD runs in two parts. The first is the web server component, the second is an RQ worker.
In development, you can run the web server through the built in debug server with manage.py
$ CONFIG="/path/to/config.conf" manage.py runserver
* Running on http://127.0.0.1:5000/
* Restarting with reloader
In production, you would want to use a different server. For instance, gunicorn with nginx in front of it. Here is an example supervisor config section.
[program:app]
environment=CONFIG="/opt/dundunduh/config.py"
directory=/opt/dundunduh/source
command=/opt/dundunduh/env/bin/gunicorn -w 4 -k gevent -b unix:/tmp/dundunduh.sock dundunduh:app
user=app
The RQ worker is always run with manage.py
$ CONFIG="/path/to/config.conf" manage.py work
09:15:09 RQ worker started, version 0.3.11
09:15:09
09:15:09 *** Listening on default...
In production, you should run several concurrently if you expect multiple jobs at once, and use the TERM signal to stop them gracefully. Here is an example supervisor config section.
[program:rq]
environment=CONFIG="/opt/dundunduh/config.py"
directory=/opt/dundunduh/source
command=/opt/dundunduh/env/bin/python /opt/dundunduh/source/manage.py work
process_name=%(program_name)s-%(process_num)s
numprocs=4
stopsignal=TERM
user=app
[group:workers]
programs=rq
Dun-Dun-Duh is licensed under MIT.
Have fun, work hard, be kind.