Trac as a stand-alone daemon (tracd)

[daniel]

Trac shouldn't have to run as a CGI. In fact it shouldn't run as a CGI at all, but rather be a stand-alone daemon (tracd), to make life easier for everyone.

#1 Goal: Ease of Installation

Installing Trac will be much easier once it doesn't depend on apache configuration and fairly complicated setup.

The goal is to have a process as:

• Install from distribution.
• Start tracd with arguments specifying what database to use (create if non-existant).
• Point web-browser to localhost:<port>, and set everything up from there.

#2: Increased Performance

Having tracd will let us set up Trac to run behind a forwarding proxy, capable of caching. Response-times will be greatly reduced due to not having to initialize Python for every request. Caching will result in dramatically fewer reqests actually coming to tracd itself.

This should be adequeate for serving even thousands of users accessing a project.

[jonas]

I'm closing this ticket now. tracd is now in the repository and almost useful.
Individual tickets will be used for bugs and missing features. These tickets
should be filed against the "tracd" component.