21:27 <Tsoots> the idea now with scribus and get text 2 (the new text
               importing api) is to create a general XML-importer
               where user can create their own schemes for importing
21:27 <Tsoots> and make it also work otherway for saving texts with
               the same scheme
21:27 <Tsoots> then we could include some general doc markup schemes
               with scribus that could be used and if a user isn't
               happy with those they could create their own
21:28 <jeffz> that's good, but I think the key to variable documents
              is being able to setup data driven logic


What would be ideal is a template designer, whereby you can create documents based on data from a data source.  eg, a text file with fixed length records, delimited text etc or from a database.

My idea is as follows, the user would do in this order:

Using scribus, you outlay the basic elements of their document.

User adds the text "#Firstname#, you are special to us." and positions it at the top of their document, adds colourful image and a blurb of text and names the object headerText

Open the template manager-> Create a new template from the current document.

Given a tree view model,

The user can defines an input widget at the top of the tree, data in the form of columns, define column number or positional fields, user names the input widget "market_research"

The user creates a widget make a reference to this positional data using a variable for. eg, user assigns 'Firstname' to column one, 'Lastname' to column two, 'FavoriteColour' to column three.  

Each object, Firstname, Lastname, FavoriteColour are assigned types, String, String and Integer.

* Current document elements are added to the tree view automatically.

The user adds an decision widget to the tree and names it 'Red'.
User sets an expression for the decision widget.  "if(FavoriteColour == Red)"

User drags and drops the text object named headerText ", You are special to us", which they created earlier into the Red decision widget and sets the colour property to Red, renaming headerText to headerTextRed for clarity.

User repeats the process for colours green, blue, yellow and pink, selecting the text object and using the clone method to create an personalised greeting for each colour.

Now the tree looks like this:

GreetingDocument-
                \
                 -market_research(DataSource widget)
                 -colourful_image(Image)
                 -blurbtext(storytext)
                 -Red(decision widget)\                                          
                                       -HeaderTextRed
                 -Green(decision widget)\
                                         -HeaderTextGreen
                 -Blue(decision widget)\
                                         -HeaderTextBlue

As you can see, the user would be able to determine how the document is formatted by using conditionals to include/exclude/tailor parts of the document.

User saves template.

Now the user has a set of rules which can be applied to their       input data to generate Scribus documents. 

21:32 <Tsoots> and there are already plans to make scribus working as
               a headless doc server

Now that we have a template to say how to create documents, a higher level process needs to be defined.

eg, a task - when do we do this
           - how do we check for input
           - define output processing (split and sort postscript using a rules based on input data, eg. where FirstName[0] = range(a-k) split into one stream)
           - define output processing for pdf/tiff/etc, 1 stream/file per document (eg, for electronic mailing)
 


Now this is just a rough idea of what I think would be good to see from powerful DTP application, I'm pretty sure it makes sense.
                 
                 
ps. I hope this makes sense.