Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [elk-dev] The new ElkGraph
Hi,

just a thought that I had yesterday: once we have the new graph and its textual (Xtext based) format in place, we could make a testing infrastructure that allows to write very compact tests using Xtend. I sketched an example to illustrate the idea:

@Test
def void testSimpleLayering() {
    val graph = parse(‘''
        algorithm = org.eclipse.elk.layered
        direction = right
        node n1
        node n2
        edge n1 -> n2
    ''')
    graph.layout()
    val n1 = graph.getNode('n1')
    val n2 = graph.getNode('n2')
    assertTrue(n1.x < n2.x)
}

The utility method parse would use the Xtext / Antlr parser to get an ElkGraph instance. The extension method layout would invoke the graph layout engine. Using getNode one could retrieve the node with a certain ID. As an alternative, this should also work:

    val graph = parse('''
        node n1
        node n2
        edge n1 -> n2
    ''')
    graph.layout(#{
        CoreOptions.ALGORITHM -> 'org.eclipse.elk.layered',
        CoreOptions.DIRECTION -> Direction.RIGHT
    })

Here the argument of layout is a map literal using the Xtend syntax. The given option mapping would be applied to all appropriate elements in the graph prior to invoking the layout.

This could help in building a good test base, which I believe is one of the most important things the ELK needs at the moment.

Miro


On 15 Nov 2016, at 16:36, Christoph Daniel Schulze <cds@xxxxxxxxxxxxxxxxxxxxxx> wrote:

Greeting, fellow ELK developers!

The new ELkGraph is pretty much done and everything is migrated (see the
cds/realelkgraph branch). "Everything is migrated" in this context means
"everything compiles". I do expect the results to be full of strange
coordinate conversion bugs, which is why I have not merged the branch
onto master yet.

Before we can track down the bugs, we need something to track them down
with. This is why my attention will now shift to KIELER Pragmatics: once
KLighD works with the new graph, we can start finding and fixing bugs. I
expect KLighD to work with the new ELK Graph by the end of the week, so
I guess we can start fixing bugs starting next week.

Once everything has settled down a bit, there are two things left to be
done:

1. Migrate some of the basic math data structures to the elk.graph plug-in.

2. Investigate if we can get rid of persistent entries. I expect that we
need to give the graph plug-in access to some sort of a mechanism for
properly deserialising layout options, something that we currently do in
the core ELK plug-in.

Comments and suggestions welcome!

Cheers,
Christoph Daniel


_______________________________________________
elk-dev mailing list
elk-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/elk-dev

--
Dr. Miro Spönemann
Software engineer and consultant

TypeFox GmbH
Am Germaniahafen 1, 24143 Kiel

Tel.: +49 151 42679459

Sitz: Kiel, Registergericht: Amtsgericht Kiel, HRB 17385
Geschäftsführer: Sven Efftinge, Dr. Jan Köhnlein, Moritz Eysholdt

Back to the top