fisterra-devel

[Fisterra-devel] Document analysis

[Sergio V. S. <sv...@ig...>]

Hi,

the Fisterra team is analyzing the document pattern that will be used in
the next release of Distribution project. This analysis pattern is used
for managing all actions related to documents like tickets, invoices,
orders... 

We will evaluate also the insertion of some kind of workflow.

These are the requirements I propose. I would like to encourage
everybody to contribute to this process.

     * There are some document types and a document can be converted
       from a set of origin types to a set of destination types.
      * A document with a given type can be in different states.
      * A document with a given state can have different versions.
      * We must be able to browse the life cicle of a given document,
        i.e, all its ancestors.
      * Document transitions: a document should be transformed between
        two types with the following conditions 
              * a transition is made of one or more actions
              * there must be almost one reponsible for each action, who
                will validate the action. The responsible could delegate
                its decission in another actor.
              * the action could have some constraints, for example time
                constraints, localization constraints.
              * the actions and even the transitions could have some
                kind of priority.
              * the flow could be from multiple documents to a single or
                even multiple documents too. For example we could create
                an invoice from a set of tickes. Another sample is
                flowing from a single ticket to a refund that will
                generate also a payment and an incident.

Bye.

[Fisterra-devel] Document analysis

[Sergio V. S. <sv...@ig...>]

Hi,

following with the document analysis I had made an initial design for a
document workflow. You can found it as attachment of this mail.

There is another attachment with a typical workflow design. My design
was based on it.

We should discuss the implementation of this design later.

Bye.

Re: [Fisterra-devel] Document analysis

[Sergio V. S. <sv...@ig...>]

Hi,

There are some things that I forgot in my first mail.
   * there could be some document versions but only the last one should
be active, i.e, the other ones souldn't be taken into account for
business processes (in a production environment I mean, this does not
include other processes like data warehouse).
   * the date and the one who performed the document transition should
be registered.
   * a document could be identified with:
      * a code sequentially generated which is document type depending
      * a barcode which is document type depending. This is very useful
for reading documents with barcode readers.

I have attached the diagram that represents the data structures into a
RDBMS.

There were no suggestions :( for my last mail, I hope almost one for
this one.

Bye.

Re: [Fisterra-devel] Document analysis

[Alejandro C. <ac...@ig...>]

On Tue, Apr 26, 2005 at 12:52:38PM +0200, Sergio Villar Senin wrote:
> Hi,
> 
> There are some things that I forgot in my first mail.
>    * there could be some document versions but only the last one should
> be active, i.e, the other ones souldn't be taken into account for
> business processes (in a production environment I mean, this does not
> include other processes like data warehouse).

What would it happen if we want to recover an old release of a document?

>    * the date and the one who performed the document transition should
> be registered.
>    * a document could be identified with:
>       * a code sequentially generated which is document type depending
>       * a barcode which is document type depending. This is very useful
> for reading documents with barcode readers.
>

Why do we need two codes? Could we use just the barcode?

-- 

Alejandro Garcia Castro
mailto: ac...@ig...
http://www.igalia.com

Re: [Fisterra-devel] Document analysis

[Sergio V. S. <sv...@ig...>]

On Mar, 2005-04-26 at 13:04 +0200, Alejandro Garc=EDa Castro wrote:

> What would it happen if we want to recover an old release of a document?
As you can see in the picture, there is a relationship between a
document and its  previous version.

> Why do we need two codes? Could we use just the barcode?
Well it is a good question, I will justify my decission, the reason for
using two identificators is that the barcode is used for physical
document management, I mean, printed documents could be easily imported
to the information system with a simple barcode reader.=20

The serial code is a document-recipient-oriented identifier. Some
providers could demand us to issue their invoices with a particular
numeration. So I thought it was a property of invoices and not of all
documents. But the same requirements are met by delivery notes or even
tickets for some kind of customers (big enterprises). That is the reason
why I think it is a property of the document.

What do you think?

Bye.