While developing expressive.js, I ran into an issue where my development's express server was running on port 3000 but my production environment's server was running on port 4000. For that reason, among others, I needed to create a way to allow my application to have some custom configuration. I ended up settling with the following solution.
The idea is that I have two configuration json files in /config. config.json
is for server-specific configuration. For example, the port that expressjs
listens on, or my redis username and password, or my Twitter api token.
siteconfig.json is for public-facing configuration. For example, the site's
name or custom footer content.
My config.js file will read the two JSON files, parse them and export them as
config.
I'll load my config into my app with var config = require('./config/config');
and my configuration will be accessible via config and my site configuration
will be accessed via config.site.
I also pass my site configuration to my views with [app.locals]:
app.configure(function() {
...
app.locals.config = config.site;
});
Coupled along with the fact that my .gitignore ignores all files in /config,
the public repository will not contain any sensitive/user-specific information.