First I’ll say that everyone I’ve shown this graph thinks it’s really cool, and it is far more helpful than the large yml file it was created from. So thanks for this.
Having said that, from this example graph, you can’t always tell where the lines terminate.For some lines it’s clear that they go under a given box, but in other cases It’s difficult to tell if they go under or if one line is terminating on one side of the box, and then another line on the other side of the box is a different line. Make sense?
Anyway, what improvements are you considering here, if any?
The rendering is eventually probably going to be rewritten with a different graphing library because it’s not easy to make the elements interactive with this one (so you can have actions/links on the nodes) and (stupid as this sounds) I haven’t been able to get it to show arrow heads on the edges correctly.
The node layout and routing of the edges can be tweaked but is a fairly complex problem. We will probably try to improve it and then make the nodes draggable so you can push things around as needed
Is this a real stack of an actual thing you’d use? If so could you give us an anonymized version of the yml so we can use it for testing? It’s hard to come up with more complicated but realistic examples of stuff like this. You can strip it down as necessary to almost nothing but the links… The images can all be busybox or something simple, and you can make the names all ‘svcA’, ‘B’, etc if necessary since you’re blocking them out here. Though realistic names would make it easier to look at & understand. It doesn’t have to be made public if you don’t want.
We should improve our native graph, but outputting dot is an interesting idea. What if that was available only from the command line? Such a thing should be simple enough to add to rancher-compose.