I'd like to propose a couple of minor modifications to the current spec : KatieParlante20050624
Those modifications will not modify the spirit of the spec and will make implementation easier.

Those are:

• Add a # character to the description row: this will make the code resilient to typos in those lines. It will also allow testers to add comments to the doc.
• Kind appear in each line: this will make parsing easier without burdening the user. It will also allow for items to be listed out of order without breaking the code.
• Reference by displayName: this was listed as a "nice to have" in the spec but it turns out it's easier to implement that way in Python (thanks Python!... :))
• If reference name doesn't exist a new item will be created: was also listed in the "nice to have" and easier to implement than error handling.
• Add a startTime parameter in Note, Task, CalendarEvent, EventTask
• Mail message item available

The translation of those changes in the spec example matrix are in red in the following table:
collection and email separator = ';'

#Collection	displayName
Collection	Home
Collection	Work
#Note	displayName	collection	creationDate	creationTime
Note	reading list	Home; Work	1/11/2005
Note	restauraunt recommendations	Home	5/5/2005	3:25:00 PM
#Task	displayName	collection	dueDate	dueTime
Task	read paper	Home	6/25/2005
#CalendarEvent	displayName	collection	startDate	startTime	anyTime	allDay	duration	reminderTime	location	importance
CalendarEvent	party		7/4/2005	1:30:00 AM		TRUE	120		PARIS
#EventTask	displayName	collection	startDate	startTime	anyTime	allDay	duration	reminderTime	location	importance
EventTask	meeting	NewCollection; Work	6/30/2005	*			*		Office
#MailMessage	subject	collection	dateSent	timeSent	fromAdress	toAdress	outgoing	stampEvent	stampTask	body
MailMessage	How are you?	Message	7/15/2005	4:00:00 PM	foo@tralala.org	fifi@bar.com; zaza@tutu.fr	TRUE			Hello, How are you fifi...

Notes on implementation :

• The python script will parse a CSV file (generated from a spreadsheet tool)
• When the token (*) is used for a field the script generate semi-random data for the field
• If a field is not filled a default value is used
• US and EU hour format are allowed

How to invoque the script :

• From the "Generate data from a file" dialog invoked from the "Test" menu
• From the command line using the --createData parameter: ~> release/RunChandler --createData=/User/Home/file.csv

About CSV file format :

• choose field separator = ','
• an example of Excel file : data.xls
• This script is tested to work with the Excel CSV (comma separated) export
• With OpenOffice CSV export choose nothing for text delimiter