We're trying a different architecture on GAE.
This project is licensed under the Apache License, version 2.
If you know a thing or two about computers and youre're interested in contributing, check out https://github.com/MayOneUS/wiki/wiki
The majority of the site is just basic static content, as it should be. That markup lives in "markup/" in the form of jade files. If you don't know what those are, they're very simple, and remove literally 50% of HTML's boilerplate so they're worth it. Read the 3 minute tutorial.
Stylesheets are in "stylesheets/" as sass files. See last paragraph for why that's good.
Ideally there will be little enough JS that no framework will be necessary (fingers crossed).
The backend will be very simple with two endpoints
- Pledge. This has to be done in coordination with stripe so that stripe stores the credit card info, and we only store an opaque token and the pledge amount. This will write to what'll be probably the only table in the datastore.
- GetTotal: Simple sum over pledges. Store it in memcache, expire every few minutes. Boom.
- Python App Engine SDK
- NPM
- sass (with ruby installed,
gem install sass)
After checking out the code, run npm install. To start the server, run npm start and go to
http://localhost:8080. That's it!
If you want to rapidly set up a development environment that already has npm and dependencies installed using docker, try running
docker run -t -i -v /path/to/checkout:/develop jtolds/mayone-gae /bin/bash
Then in that new shell
stunnel4 /etc/stunnel/https.conf
cd /develop
npm install
npm start
After running npm install and grunt local, copy build/config.json over to backend/ and run
python testrunner.py in the project's root directory.
We have 4 deployment environments available, all of which can be set up with grunt (installed by npm).
- local: For normal development, with code updates on every reload. Run
npm startor equivalentlynode_modules/.bin/grunt local.- Note that in local mode, you can't actually send a transaction to stripe due to an SSL bug in dev_appserver. But you can get right up to that point before it fails, which is generally good enough.
- dev: This is an independant instance of the app running at https://pure-spring-568.appspot.com. We can do
whatever we want here because the data's all fake. It also uses Stripe's test keys, so feel free to submit test
transactions with credit card 4242 4242 4242 4242. To deploy, run
./node_modules/.bin/grunt dev, and thenappcfg.py --oauth2 update build/. When testing, it works best to hit this URL: https://pledgedev.mayday.us/pledge. - staging: This is the real app, but a separate version that's available at https://staging-dot-mayday-pac.appspot.com/.
Notably, it uses the same datastore as prod, and the real stripe keys. Any code that touches
the data should be deployed here with caution, but if not, you can be a little lax since real
traffic doesn't hit it. To deploy, run
./node_modules/.bin/grunt staging, and thenappcfg.py --oauth2 update build/. - prod: The real McCoy. Don't break it. To deploy, run
./node_modules/.bin/grunt prod, and thenappcfg.py --oauth2 update build/.
We don't have much in the way of release procedures, so it just comes down to how paranoid you want to be. Some changes can go straight to prod. Some should go through dev first. Some through both dev and staging before prod. App Engine even allows you do to traffic splitting, so we could theoretically send N% of queries to staging if we're really paranoid about something. Use good judgement and run it by team memebers if you're unsure.
If you're pushing to prod, please tag the commit you're pushing with the next available "v1,v2,v3,..." tag, and remember to git push --tags. That way if we need to roll back we have a sense of what used to be in production.
If npm start generates an out of files error and prints a lot of messages like this:
Warning: EMFILE: Too many opened files.
You probably have an orphan app engine process running. Try looking for and killing any child processes. Worst case, you may need to increase your inotify maximums; you can do that on a Linux system with this command:
echo 1024 > /proc/sys/fs/inotify/max_user_instances