Re: Border floating connections

[Lars Clausen <lrclause cs uiuc edu>]

On 17 Sep 2003, Wynand Winterbach wrote:
> I started using DIA for graph theory related figures, and it's helped a
> lot.
>
> However, there is one feature that I've always longed for. I'll explain
> as follows:
>
> the edges that are drawn should preferably be connected to the center of
> a vertex, and this is easily achieved by defining a custom shape;
> however, this only works for when the lines don't have arrow tips on
> them, so of course what does happen when we do have arrows, is a
> situation as follows (where the O's represent vertices, and there is a
> line from left to right)...
>
>       \
> O------0
>       /
>
> instead of
>
>      \
> O----->0
>      /
>
> The solution is that either:
>
> a) the edges still get connected to the center of the vertex, but a line
> must be coded which "knows" where the object that it is pointing to has
> its border, so that it can correctly draw the arrow tip.
>
> b) the whole border of a vertex should be able to serve as a connection
> point - this would mean that the exact point of connection would move
> smoothly along the border as the other side of the line gets moved.
>
> Is there any way to implement this currently? If so, I'd like to give it
> a bash.

There's actually some code in there that does that.  Take a look at
objects/standard/line.c, it has some commented out sections about 'line
gaps'.  It essentially allows a gap between the connection point and the
drawn end of the line, specified either as a absolute distance, a
percentage, or just as 'to the edge'.  I haven't set that as generally
available, since the interface is rough and fairly incomplete, but if you
want to play with it -- possibly make it work with other connectors than
lines -- that'd be great.  

I'm thinking the main use of it will be to allow exactly what you describe
in a), that the middle of an object (or the entire object) is a connection
point, but the connector just automatically stops at the end.  The
possibility of having other kinds of gaps could be useful sometimes, but is
not as essential.

-Lars

-- 
Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause)| HÃrdgrim 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?