I started learning Ruby in part so I could do some web scripting, and 
have already twice tried installing Rails, with my usual catastrophic 
installer-malfunction-experiences with Ruby-related material. (I like 
Ruby, but I *hate* having to install it.) At the moment I have a brand 
new XServe with a nice fresh install of stuff. I have not yet tortured 
it (and myself) with Ruby or Rails yet. Ruby's just on my laptop for 
now, and Rails isn't going on anything until I'm ready.

One of the "getting ready" things is rebuilding and repopulating a 
PostgreSQL database. As it happens, I can afford to redesign the 
schema, since we're retiring the current apps that use it. Now, I've 
seen bits and pieces of some very specific expectations that Rails has 
for databases.

So why is the complete description a Rails-conformant database a giant 
secret? I've Googled all over the web trying to find what, logically, 
ought to be right up front in big bold type on the rubyonrails.org home 
page: how to name/design a database so Rails "gets it."

The only thing I've found so far that even comes close was in a PDF 
file, specifically Jeffrey Hick's "Many to Many" tutorial. Um? That's 
preposterously obscure, or so it seems to me.

So far, I've got that it wants my table names to be plural, that trying 
to use anything other than an autoincrementing integer named "id" for 
the identity column is likely to be a real PITA, that foreign keys have 
to be [tablename][underscore]"id", and (a secret exclusively in Hick's 
article), linking tables must be named after the two tables to which 
they link, and in alphabetical order!

So what other secrets should I have, and why isn't this the very first 
article in the database-related Wiki articles or otherwise located so 
that I'm tripping over it, instead of scratching my head?

Lugubriously,
a not-easily-dissuaded Ruby newbie.