Mr von Waldow,
Well, there are a couple of ways this can be done. One is to try
it graphically, presenting the usually class hierarchy for the
objects in the system that aren't in the system class libraries.
The other portrait that's sometimes useful is a so-called
"instantiation graph", showing an arc from class A to class B
if a message in A instantiates an object of class B. This can
be limited to the non-system classes, or go all the way and
show the system classes, too.
The other possibility is to describe the system as an interlinked
assembly of the standard patterns, although juggling or refactoring
things might be necessary to do all these recognitions.
All these ways might be good ways to analyze and document, and
without knowing more about the system I'd have to recommend them
all.
Hope this helps,
--jtg
Wulf von Waldow wrote:
>
> Hi everyone, I've started doing object oriented analysis of a large already
> finished system. I was hoping that there might be some advice on what would
> be the best way to do this.. top down.. bottom up.. ,..and things like how
> can I keep it tidy so that my diagrams don't look like spagitti also if
> anyone knows any good books or information available it would be great. thx
[snip]
--
-----------------------------------------------------------------------
Jan Theodore Galkowski ºoº
[hidden email]
algebraist.com/ wikiweb.com/ atlasti.de/
[hidden email]
***********************************************************************
Ask me about the Disney Vacation Club, or see algebraist.com/dvc.htm
***********************************************************************
"Smalltalk? Yes, it's really that slick."
-----------------------------------------------------------------------
Want to know more? Check out whysmalltalk.com/, object-arts.com/
***********************************************************************
Locked
