The first thing you'll want to do is to get an idea of the amount of data in the database. These queries get general information about the major data sources in the database:

From this it should be very obvious that the data contained in the database are not the complete record of everything that ever happened in the GNOME community. The data are focused around questions that I found most interesting and was focusing on for my own research. Thus, not all projects have equal amounts of information in the database.

This query shows how much of the data is linked up. We first take an email address (email_address) and map that to person (person), which can be mapped to a cvs account (users), which may have created multiple cvs commits (cvs_commit), on multiple projects (project), which belong to the large evolution master project (master_project). I realize that it is possible to compress this query some by eliminating the person table and various other optimizations, but it's left fully verbose here to make the process clearer.

[and several hundred more]

This is an expansion of the previous query where we continue the join two steps further to join find the all of the mailing list posts that the email_address has done (mail_message) and then figure out which lists those posts were done on (mailing_list).

The blog engine was something I wrote for a one off weekend project. From there it just ran for a couple of years. That's the power of cron. The basic premise was that at some points I crawled the various blog planets for their OPML lists of subscriptions. I then visited all of those URLs religiously. The planet lists were not frequently updated, so this doesn't represent and entirely true image, but it might be interesting to natural language people or something like that.