[Uml-devel] new omf

[Andrew Sutton]

all,

just thought i'd let you all know... i've got a new and improved version of 
the omf out (0.96) along with some cool little programs that do stuff with 
it.

this version basically fixes some uml-specific problems i've been having (like 
ambiguous feature lookups and duplicate associations) rather than adding new 
features. there are two "new" features: one is a redesigned XMI reader that 
is actually composed of two XMI parsers (one for the MOF) and one for 
everything else. It turns out you can't _really_ make a single parser that 
generic, although god knows, i tried. the other new feature is kind of... 
well, not that exciting. i started implementing some of the methods defined 
for MOF objects. It turns out to be useful to have some of them while parsing 
models.

i'v also made a little change to kmodelview... when you click on an object in 
the tree it will show attributes and references (it basically grabs the name 
of anything being pointed to). it's actually kind of nice. oh... right. and 
my model iterator continues to work, although i'm certain it's fundamentally 
broken somewhere (yeah me!)

i've got two new programs worth mentioning...

the first is mmgen. this is the bastard i've been using to generate the mof 
and uml implementations. i hate this program, but it's been useful. it might 
be worth looking at if you're interested...

the other is actually pretty cool. it's basically a program that converts 
umbrello XMI files to OMF XMI files - it doesn't handle ALL of umbrello's 
stuff yet, but it's a decent start. It successfully translates packages, 
classes, attributes, operations and parameters. it only really works with the 
most recent versions of umbrello (although i've only tested on the 1.2 beta 
built from kde 3.1.93). unless there are significant changes to the listed 
functions, it should still work.

let's see... anything else? oh right. my programmer's guide/design doc/soapbox 
is available too. so far it's got a walkthru of two simple UML programs and a 
fairly lengthy discussion of the OMF type library. this is very much a work 
in progress and will (one day, i hope) cover everything from basic omf 
programming to specific uml issues (like stereotypes and templates) to 
advanced features like building extension attributes, and creating extension 
metamodels.

as always - or at least until i stopped trusting the kent systems team - all 
this stuff can be found here:

http://warhol.cs.kent.edu/~ansutton/omf

andy