[Uml-devel] good news - maybe

[Andrew Sutton]

hey list,

i've got some good news. i've managed to convince my cs department that they 
should give me an indepndent study working on umbrello :) however, because 
its department work, i have to focus some of the attentions of my work on 
subjects of their choosing. all my work is also going to focus on the next 
version because i don't think i can fit everything into the current version. 
i'll get into that in a minute.

since classes start monday, my guess is that i'm going to have to start 
producing relatively soon. sebastian and i have been doing some of the design 
work offlist and getting the general architecture down and some of the 
stickier issues of serialization, persistence and reflection. ugh. there's 
also been some discussion of the gui requirements and design.

what i'd like to know (i guess) is how we can proceed. it seems to me that 
finishing the 1.1 release is a priority. to this end i would suggest that we 
identify any major showstoppers and focus attention on those - just to get it 
out the door. once done, 1.1 should move into a maintenance/bugfix branch and 
most of the development should focus on 2.0.

that means we'll need a 2.0 branch pretty soon. i've been working on the 2.0 
design and have parts of it completed (not nearly everything) to contribute. 
unfortunately, the new design doesn't look anything like the old design. its 
highly modularized and is (or will be) based on a plugin architecture. almost 
everything outside of the core library and the gui itself will be a plugin of 
some sort (probably not kparts) - each contributing functionality to the 
application. also, some of the material (as we know from previous discussion) 
is really complex and abstract and takes some time to wrap your head around. 
i think i've got a pretty good handle on that stuff so i'll be able to answer 
some of the questions.

now, because i'm getting credit for this, i have to think about my departments 
interests. there's quite a few. here's a brief list of things that are going 
to be taken into account during design and might start getting some code 
attention.
	- uml extensions for aspect-oriented programming
	- automated layout and aesthetics engines for diagrams
	- a framework for a design metrics suite
	- a framework for statistical analysis of design
	- codegen and import for srcML (XML source code)
	- exports for 3d visualization tools
	- automated identification of boundary, control and entity classes/objects 
within a design
	- automated identification of concerns within a design

i think there may have been some other suggestions, but that's what i remember 
being mentioned. however, i don't want these to interfere with the creation 
of the actual application. as far as i'm concerned, i just have to allocate 
space in the design for these things. i believe that most of them can be 
built as plugins to the design that sebastian and i have been discussing. who 
knows, maybe i'll even get some of the grad students to contribute as part of 
there theses or their own independant studies.

anyway, there's a lot of things that we can do to make this the best CASE tool 
around - and that's what i'd like to aim for.

so, given this, how do i/we proceed? who is interested in working on 2.0? are 
there any problems with what i'm proposing that i'm not aware of?

andy