[RFC] SRS Section 3

James Robertson jwrober at linuxfromscratch.org
Thu Feb 10 04:30:31 PST 2005


Matthew Burgess wrote:

>> The IBM Rational Unified Process calls the low level design documents 
>> different things, use cases mostly.
> 
> IBM call UC's *low* level design?  From my experience they've been about 
> has high-level as you can get, but then my experience is very limited 
> when it comes to use cases.
> 
The phrase "Use Case" has many meanings to the IBM RUP.  At the top most 
level (that would be high) you have a Use Case diagram.  In the diagram 
you lay out your Actors and Use Case objects to show the relationships 
between them using UML (unified modeling language).  In UML, an actor is 
represented as a stick figure and a UC is an oval. Each UC "oval" on the 
top level diagram is then ripped apart to a more granular level to 
describe the full specification of the UC (in separate docs).  At this 
level you discuss the flow of "work" of the actor(s) interacting with 
the system in the UC.  I was thinking that we could look at writing the 
UC spec sheets in XML Docbook.  I have a template in MS Word that I can 
port to a quick wiki page for us to use if you are interested.  You can 
go even lower than this using UML and the RUP, but it is WAY overkill 
for this little project.  I think the top two layers I described here 
will be a good exercise for all of us.  What makes it nice is everyone 
can *read* what the tool will do in a very logical manner and can 
imaging using it in that way.  That way we all can get consensus of what 
it will do when written.

I know there are some coders out there ready and willing to jump right 
in.  I think that is still possible once we get some high level stuff 
worked out.

Matt - I know of no OSS tools that do UML modeling.  I use MS Visio or 
IBM Rational Rose which are both commercial apps.  Can DocBook handle 
simple diagrams?  If so, we could probably do that or I can draw in 
Visio what we want and save it to html.

James



More information about the alfs-discuss mailing list