PyCon Program Committee Web App

WARNING: This is forked from njl/progcom by @njl and customized for PyCon Canada 2018.

The goal of this app is to provide a useful tool for asynchronous review of PyCon talk submissions. The program committee's job is extensive and daunting, and I'm trying to knock together a simple web app to allow the work to proceed as efficiently and effectively as I can. Requiring large groupings of busy professionals to come together at the same time to chat on IRC is hard, and doesn't feel very scalable. This is my first step toward understanding how to scale the whole thing.

Instructions for PyCon Canada 2018

The setup heavily uses Docker as a tool of choice.

Getting Started

1. Install Docker
2. Clone this repository.
3. Run the migration by executing make dev-db migration-pause migration. (DB Password is test)
4. Run docker-compose -f docker/dev.yml up -d app to start the service.

At the point, you can access to the app at http://localhost:4000/.

To reset everything (data), run make dev-db-reset.

Customizations

• Disable any lines that trigger email delivery via SendGrid.
• For daily report, we use Slack notification (webhook) instead.
• The user approval process can be done manually by running: update users set approved_on = CURRENT_TIMESTAMP where email = :email with psql or semi-auto by executing a script.
• Update pull_updates.py to decouple from the APIs from us.pycon.org as we are integrating with Papercall (with djangocon/papercall-api-import from @djangocon). (ETL)

For the ETL process, we write a script named etl_papercall_to_db.py where it borrows some parts of both djangocon/papercall-api-import and pull_updates.py.

To Run the Papercall-to-DB script

1. Log into the server. (For dev, run docker-compose -f docker/dev.yml exec app bash or SSH to the server directly.)
2. Run envdir docker-config python etl_papercall_to_db.py PAPERCALL_API_KEY.

You will need an API key (PAPERCALL_API_KEY) generated by Papercall for your own account (apparently available for any account with any level of access).

Original Documentations

Configuring the Applications

The application picks up configuration from environment variables. I like to use the envdir tool, but you can set them however you like. A complete set of configuration values, reasonable for testing, are available in dev-config/. In production, you should add a SENTRY_DSN.

You can install envdir via brew install daemontools on OS X, and apt-get install daemontools on Ubuntu and Debian.

As configured by the values in dev-config, the application connects to a local postgresql database, with username, password, and database name 'test'.

Configuring the Database

The application uses a Postgresql database. If you're not familiar with setting up Postgresql, I've included setup_db.sql for you. Getting to the point where you're able to execute those commands is going to depend on your system. If you're on a Ubuntu-like system, and you've installed postgresql via something like apt-get install postgresql, you can probably run the psql command via something like sudo -u postgres psql. On OSX, if you've installed postgresql via brew, with something like brew install postgresql, you can probably just type psql.

You can create the test database and test user via psql template1 < setup_db.sql.

The unit tests will create the tables for you, or you can do something like psql -U test test < tables.sql to create empty tables from scratch.

Running the Application

Make a virtualenv, pip install -r requirements.pip. Run the application locally via envdir dev-config ./app.py, run the tests via envdir dev-config py.test.

You can fill the database up with lots of lorem ipsum nonsense by running the script envdir dev-config ./fill_db_with_fakes.py. You can then log in with an email from the sequence user{0-24}@example.com, and a password of abc123. [email protected] is an administrator.

Deployment

You'll need deploy-config in your root directory, which should have all the appropriate secrets. From the application's root directory, you can run ansible-playbook -i hosts deploy.yaml.

Understanding The PyCon Talk Review Process

The process runs in two rounds; the first is called "screening", and is basically about winnowing out talks. Talks which aren't relevant for PyCon, have poorly prepared proposals, or otherwise won't make the cut, get eliminated from consideration early. Talks aren't compared to one another; a low-ish bar is set, and talks that don't make it over the bar are removed.

The second part of the process is "batch". In batch, talks are moved into groups, and those groups are then reviewed one at a time, with a winner or two picked from every group. Some groups feel weak enough that no winners are picked.

To turn on Batch, echo 1 > dev-config/THIS_IS_BATCH.

To disable feedback in screening, echo 1 > dev-config/CUTOFF_FEEDBACK.