[Uml-devel] kdesdk/umbrello/umbrello

[Brian Thomas]

On Thursday 05 February 2004 12:25 pm, Jonathan Riddell wrote:
> On Thu, Feb 05, 2004 at 05:59:13PM +0100, Brian Thomas wrote:
> > CVS commit by thomas:
> >
> > association/code generation patch: make them both play well together
> > again. Associations now are deleted (again) when last parent widget
> > disappears, they correctly connect to roles/classes; code generation
> > correctly tracks connections/disconnections for 'owned' objects, in
> > particular in codeclassfields and associated methods. This patch
> > probably breaks association cut and paste; but that should be
> > implemented as XMI serialization rather than copy/pasting of
> > objects, as it currently is done. Other than this, overall,
> > stability is vastly improved for the program overall, and code
> > generation/associations in particular
>
> Gosh, well done (assuming it works, I'm still compiling :)

	Heh, yes. Please, everyone, try it out. I expect to have to 
	iterate on this a little bit to remove all the kinks/bugs.

>
> Do you think this should be backported to 3.2, i.e. are you pretty
> certain it fixes problems and doesn't create any new ones?

	Yes, definitely. 

	I've been abit out of the loop recently, but I recall many email on
	what should be done for Umbrello 1.3. Thats well and good, and
	I have my own list of features I'd like to see, but I think that we havent
	really finished the work on 1.2 yet, especially given that it really 
	cant be used for much without crashing it right now.

	So I'd advocate that we have 'feature freeze' on umbrello for at
	least another month or 2 until we can wittle down the bugs in the
	3.2 version. 

	At the least, this will be my goal, unless a blizzard of semi-random feature 
	coding  starts up again, in which case, I will probably drop out from the project.	
	It just gets too exhausting for me to maintain my own bits of code, and
	fix bugs from it, when others are tossing in features to the code base.

	So, what am I really getting at? Well, I'd like to see a 'plan' for the 1.3
	design. I'd like to see us stick to that plan, feature wise, and I'd like to
	see people with new features have to *justify* why we need them. No
	more "I have a kde cvs account, so I guess I can add X to the program"
	(or, at least, no more of that for major changes to the code).

	So, in summary, I'd like to see the next few months devoted to actually
	fixing bugs and to creating a structured plan for the 1.3 design. We 
	can, of course, modify this plan over time, but in general, I'd like to see
	that the plan we create with is largely the blueprint of the work we will do
	over the next period of development.

>
> Any idea which bugs.kde.org entries this closes?

	Nope. But in the next week (Im out of umbrello time this week) I'll try to settle
	down with bugs.kde.org and figure it out. 

	regards,

	=b.t.



> Thanks
>
> Jonathan