In the deployment folder a file called "changesets" exists. In this file a list of directories exists. All these directories exists in the changesets folder.

For each directory in this file a number of sql files exists. Whenever the update script runs it will run all sql scripts in order (see naming below) in each folder listed in the changesets.txt file.

When a folder has been run once, it will not be run again in any future runs of the script.

This means that when a changeset (folder) has been run, it will do no good adding a sql file to that folder, since it will not be run again.

The files in a changeset folder are run in alphanumeric order, meaning that files beginning with 1,2,3, etc are run first. Then files beginning with a,b,c .. will be run

So this folder structure could be one way of doing it

|-changesets
|  |-AddColumnToTable
|  |  \--1 AddColumnToTable.sql
|  \-CreateTable
|     |-1 CreateTable.sql
|     \-2 AddIndices.sql
\-changesets.txt

The changesets.txt would simply look like this:
CreateTable
AddColumnToTable

The script keeps track of which files have been run in a table called db_changes. If this table does not exist when the script is run, the script will create it.

In the table, only changesets ie. directory names are saved.