If it were possible to automatically select and zoom in on a node simply by adding a command-line flag, this would lead to dramatic implications for the usability of the "go to URL" function, at what seems like a minimal cost.
I imagine a similar syntax:
java -jar yed.jar mygraphfile.graphml --focus aNodeName (or other type of node identifier)
This would simply open mygraphfile.graphml in yEd and immediately home in on the node with "aNodeName" in (or as) the label, zooming if necessary for a clear view of the node.
This would allow different yEd graph files to link to each other on a node-by-node basis, simply by entering a command with the above syntax into the "URL" field of any node, using an OS-native shortcut as a proxy, if necessary. This might be helpful when, for instance:
- A user is working with a complex graph-structured data set, but for reasons of convenience or performance would prefer to structure this data into several smaller files rather than pool everything into a single huge one. This feature would allow them to do so, while still allowing for arbitrary interconnectedness between different nodes in the involved files.
- A complex data set contains a finite number of different types of nodes. As above, putting everything into a single file might make things like layout and navigation, needlessly complicated. In this case, the user might instead consider using a separate file for each type of node, but would not be prevented from creating relationships between nodes of different types.
- The pattern might work best for complex graphs with a strongly componentized structure, with many clusters of tightly interconnected nodes with few edges interconnecting the clusters. Putting each cluster in a separate file and then interconnecting them via this or similar mechanism might dramatically help ease working with such data sets. It would also allow better version control of data sets that are built or edited by a collaborative effort, seeing as it would reduce the risk of edit conflicts arising, by separating the data into several smaller files.
For all intents and purposes, such a "URL" nodelink would serve just like any other kind of edge, except for its connecting to a node in a different file, which a purely technical distinction, not a logical structural one. I know of certain other graph or mind map editors that explicitly support such node interlinking between files. SimpleMind for Android comes to mind. yEd would already support the pattern, if only this kind of commandline call was possible.