|
|
|
|
|
|
|
Re: Layout & Routing - Any help will be appreciated [message #193264 is a reply to message #193240] |
Wed, 24 August 2005 14:46  |
Eclipse User |
|
|
|
Originally posted by: none.us.ibm.com
See ShortestPathConnectionRouter.
"Abul Bashar" <bashar@infromatik.uni-bonn.de> wrote in message
news:deiegr$bjo$1@news.eclipse.org...
> The placement of the nodes & subgraphs are really good from the algorithm.
> I am thinking to develop a router which independently observes the
> obstacles (figures) and adds optimized bendpoints to the edges. Do you
> have any clue? I don't have too much time to finish this task. Thanks a
> lot.
>
>
> "Randy Hudson" <none@us.ibm.com> wrote in message
> news:dei8ii$3dt$1@news.eclipse.org...
>> The algorithm needs some tuning for compound graphics, definitely.
>> Currently, long edges are always parented by the common ancestor subgraph
>> of the nodes at each end. In some cases it would look better to route the
>> edge through the subgraph, going out the top or bottom. Instead, the
>> edges immediately get routed outside of their container. Secondly, the
>> weighting factors of nodes in each rank needs to be adjusted better for
>> the compound cases. The edges shouldn't be crossing container twice like
>> in your graph.
>>
>> "Abul Bashar" <bashar@infromatik.uni-bonn.de> wrote in message
>> news:dei7i7$1qc$1@news.eclipse.org...
>>> Thanks Randy,
>>>
>>> Yes, I have already tried the way you've suggested. It works well for
>>> simple graph, but a nested graph looks terrible. Here I've added two
>>> more images; both are produced by the same code &
>>> CompoundDirectedGraphLayout. I have added relative bendpoints from the
>>> virtual nodes attached with edges after the layout.
>>>
>>> Have I forgot some settings with CompoundDirectedGraphLyout while
>>> treating the nested one? But I wonder the other one (nonesting.png) is
>>> also layouted by the same code and it looks good.
>>>
>>
>
>
|
|
|
Powered by
FUDForum. Page generated in 0.06884 seconds