Re: combined UML diagrams

[Lars Clausen <lrclause cs uiuc edu>]

On 28 Oct 2003, saintiss@gmx.net wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi,
>
>> Take a look at select_transitively() in app/select.c, it does the same
>> kind of thing, both ways.
>>
>> You're quite right that you're getting the same thing.  The connection
>> point is what's visualized by a little blue x, the handle is the box.  A
>> handle can only be connected to one thing, a connection point can have a
>> number of handles connected to it.  So when you go from handle to
>> object, you need to look at handle->connected_to->object.
>
> Hm, according to what you're saying, one of the handles of the class
> should be connected to the same connectionpoint then, right? Well, when I
> do this:
>
> for (i = 0; i <= 7; i++) {
> handle = &((elem->resize_handles)[i]);
> conpoint = handle->connected_to;
>
> all those conpoints seem to be NULL... So something's definately wrong
> then?

No, no, the connectionpoint is on the class (the blue x).  The handles on
the class don't connect to anything.  The handles on the line connects to
the connection point on the class.  That's why there's an x in them as
well.  So take the handle from the orthconn, find the connpoint it's
connected to, find the object on that (not the connected list), and you
have the class.

>>
>> That sounds really strange, I'm sure people would have noticed by now --
>> and indeed, it doesn't do that for me.  Can you send me a diagram that
>> behaves this way?
>
> Ok, I've attached one... However, I just noticed that the behaviour only 
> occurs if the classes are parented inside a package...

Ah, yes, that I see.  Not good.

-Lars

-- 
Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause)| HĂ„rdgrim of Numenor
"I do not agree with a word that you say, but I   |----------------------------
will defend to the death your right to say it."   | Where are we going, and
    --Evelyn Beatrice Hall paraphrasing Voltaire  | what's with the handbasket?