Aspect modelling(?)

Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic

   Hello,

I'm wondering how to model some things.

In one graph I have computers from a cabling perspective.
In another graph I have the same computers, but now from an subnet perspective.

In the cabling perspective I draw my cables (works fine)
In my subnet perspective I model the subnets and the ip addresses (also works fine).

Now I need to link them together. The problem is that objects in both graphs are actually the same, but from a different perspective, and the intended users already expressed that they would like to have different visuals for them in both graphs. As far as I know that leaves me no other option (in metaedit+) than to use two objects and somehow define a one on one relationship between these objects.

What approach do you suggest?

I can have a third graph where I draw the relation between two things that are actually the same.. sort of the same anyway (not good). This doesn't feel completely ok, but at least I have to enter the information only once.

I can add an attribute to each object pointing the its counterpart in the other graph (which makes them also part of the current graph of course, but not visually). With this approach I do not need a third graph, but users need to enter the information twice (not good), and can make mistakes when assigning values to the attributes (os view object points to a different cable view object and vice versa)

Looking at the second option it seems the worst of the two at first sight, but is it really? In the first option I have to keep track of the graphs / views in which I define the link, and how do I keep track of these graphs when generating my network configuration out of it?

Regards,
Jeroen Kouwer

Author	Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic
jkouwer Members Profile Send Private Message Find Members Posts Add to Buddy List Major Contributor Joined: 08.Mar.2010 Points: 22	Post Options Post Reply Quote jkouwer Report Post Thanks(0) Quote Reply Topic: Aspect modelling(?) Posted: 25.Jun.2010 at 20:00
	Hello, I'm wondering how to model some things. In one graph I have computers from a cabling perspective. In another graph I have the same computers, but now from an subnet perspective. In the cabling perspective I draw my cables (works fine) In my subnet perspective I model the subnets and the ip addresses (also works fine). Now I need to link them together. The problem is that objects in both graphs are actually the same, but from a different perspective, and the intended users already expressed that they would like to have different visuals for them in both graphs. As far as I know that leaves me no other option (in metaedit+) than to use two objects and somehow define a one on one relationship between these objects. What approach do you suggest? I can have a third graph where I draw the relation between two things that are actually the same.. sort of the same anyway (not good). This doesn't feel completely ok, but at least I have to enter the information only once. I can add an attribute to each object pointing the its counterpart in the other graph (which makes them also part of the current graph of course, but not visually). With this approach I do not need a third graph, but users need to enter the information twice (not good), and can make mistakes when assigning values to the attributes (os view object points to a different cable view object and vice versa) Looking at the second option it seems the worst of the two at first sight, but is it really? In the first option I have to keep track of the graphs / views in which I define the link, and how do I keep track of these graphs when generating my network configuration out of it? Regards, Jeroen Kouwer

stevek Members Profile Send Private Message Find Members Posts Add to Buddy List MetaCase Joined: 11.Mar.2008 Points: 643	Post Options Post Reply Quote stevek Report Post Thanks(0) Quote Reply Posted: 26.Jun.2010 at 00:46
	Hello Jeroen, If the only difference between the objects is a different symbol, then there's a third option that may well be best. You can have just one object type, but have effectively two symbols for it. Both symbols are drawn in the same symbol editor, e.g. side by side. You then make the left half into one group and the right half into another group, and in each group's Format dialog add a Condition that makes it display only in the desired graph type. That is easily done with a Generator containing type;1 - i.e. the type of the graph, one layer of stack out from the object - compared to be equal to the name of the Graph type. You can then drag the groups so they are on top of each other, and have just one default connectable that does as good a job of possible of being a suitable outline for both. The condition will take care of displaying only the right elements in the right graph. If the objects need to be different for some reason, another approach is to have a main object type that contains the properties, and then in each graph a different simple wrapper object type with a single property, whose value is an instance of the main object type. I'd suggest you also double check with your intended users / domain experts to what extent these things really are the same. Ask them tricky questions :-), e.g. if I unplug the cable from my PC and its switch, and connect it with a different cable to a different switch, it still ends up with the same IP and subnet; conversely I can keep the same cable but change a setting in the PC or server, and end up with a different IP and even a different logical subnet. Ask them what are the kinds of changes that are likely to happen in the real world, and how that should be reflected in the models. All the best, Steve