Table of Contents
These instructions outline how to use HTSQL using a cloud provider where you have a limited application environment.
There are some basic steps that you must do if you’re not already using Heroku service. Heroku’s server management tools are written in Ruby and use Git.
If you follow Heroku’s quickstart don’t follow the remaining instructions past the Prerequisites since this is intended for ruby application developers.
Next is an application template that matches with Heroku needs. You’ll need to create an empty “project” directory with the following files & content.
HTSQL PyYAML psycopg2
web: python run_htsql.py
# shell plugin as a default command tweak.shell.default:
# create HTSQL application import os, yaml from htsql import HTSQL config = yaml.load(open("config.yaml").read()) dburi = os.environ['DATABASE_URL'].replace('postgres','pgsql') app = HTSQL(dburi, config) # start webserver from wsgiref.simple_server import make_server port = int(os.environ['PORT']) srv = make_server('0.0.0.0', port, app) srv.serve_forever()
These files can be found in the demo/heroku portion of the HTSQL distribution.
Once those files are created, deployment to Heroku’s system is done by first creating a local git repository for your project:
$ git init $ git add . $ git commit -m init
Then, initializing the heroku stack:
$ heroku create --stack cedar
At this point, a PostgreSQL database is needed. For testing, you could use a shared database:
$ heroku addons:add shared-database
Then, you’re ready to push HTSQL to Heroku:
$ git push heroku master
You could now go to the URL of the hosted service and enter /'Hello World' for a very basic test. If HTSQL server is not up and running, look at the logs:
$ heroku logs
Likely you will want to customize your deployment, say running it along side your web application and using nginx to dispatch requests to some prefix to this process.
One major limitation of Heroku deployment is that Ruby on Rails applications typically do not define foreign keys. Furthermore, the tool Heroku uses to do import and export, taps, discards any foreign key constraints when migrating schemas. Since HTSQL relies upon foreign keys to determine linking, it’s usage is limited to single table queries or manually specified links.