In Part 8 we saw how the Change Manager failed to accomplish a very simple task. To be fair, the Change Manager is pretty good and almost always is capable of analyzing simple changes. Normally it fails only when the changes are complex. In order to explain my theory I was looking for a simple case failure (like the hole + chamfer failure) and had to actually look really hard to find one.
In this part we will see how the Change Manager succeeds in giving us a decent feature tree, without any sculpt features, for the same hole + chamfer case which failed in Part 8. Lets rewind and do things a bit differently this time. I create the same cube in Inventor 2010 and get it into Inventor Fusion where I cut out a hole. But instead of adding a chamfer, I save the model to a DWG file and open it back in Inventor 2010. The change manager kicks in and sure enough it reports one change and recommends a Create Extrude treatment for creating the hole.
I click the Apply All button and a second extrusion feature gets added to the feature tree. The first extrusion was for the cube and the second for the hole. I save the Inventor model and overwrite the original IPT file. So as you can see I just completed a round trip by taking the original IPT file to Inventor Fusion and bringing a DWG version of it back to Inventor 2010 and overwriting the IPT file.
Next I open this updated IPT file in Inventor Fusion and add a chamfer. I again save a DWG and open it Inventor 2010. The Change Manager kicks in and reports a single Create Chamfer treatment.
Again I simply click the Apply All button and the Change Manager adds a chamfer feature to the feature tree.
As you can see we have two extrusions and one chamfer, no sculpt features. This is precisely the feature tree that the Change Manager failed to cook up with in Part 8. The only thing we did differently this time was that we performed a round trip after creating the hole.
This very simple example only goes to illustrate a larger systemic problem with the way the Inventor Fusion technology has been packaged. Since users have to deal with two applications (Inventor 2010 and Inventor Fusion) they end up asking the Change Manager to deal with far more changes that it can handle. This often leads to undesirable results. Sometimes sculpt features are added as a last ditch attempt to accommodate the change in the feature tree. Sometimes, even that is not possible and you have no option than to ignore the change. I believe this is huge drawback especially if the change was a required one. And then sometimes, the Change Manager simply cannot handle it and crashes, taking Inventor 2010 down with it.
Its like asking a Magician to do ten magic tricks all at the same time. What makes is worse is the fact that the output of one trick is actually the input to the next. All we are interested is in the output of the last trick and instead of letting the magician perform the tricks in a sequential manner, we are unnecessarily forcing him to do them all at once. He is bound to goof up.
If Autodesk had changed Inventor’s UI to allow direct modeling right inside Inventor 2010 itself, then after every direct modeling operation, the Change Manager could be invoked automatically in the background. It would do its magic (just one trick) and edit the feature tree. When the user performed the next direct modeling operation, the Change Manager would again kick in and tweak the feature tree a little more. Since the Change Manager has to perform one trick only, the chance of its success would be way more than if it was asked to do a bunch of tricks all at the same time. And if it failed, then the user could be asked to choose from a list of treatments just like he is asked now. Quite similar to how the user is interrupted when a build error occurs if he edits a feature that the modeling kernel cannot handle.
During my tests on real world data, the Change Manager failed numerous times. In each of these cases, I restarted my experiment and made a round trip after each individual direct modeling change in Fusion. Almost always, it worked as it was supposed to. Quite similar to how the hole + chamfer change worked above and failed when I tried to do it all at once in Part 8.
This is why I said in Part 8 and the Fusion technology is a success, but the way it has been packaged in a total disaster. The sad part is that not all people trying out this technology are going to do what I did – make round trips every time they perform a direct modeling operation in Fusion. They will open their IPT file in Fusion, thrash it to their hearts content, get it back into Inventor 2010 and the Change Manager will most probably fail. In fact, I am surprised that it actually works when it does. It is quite clear to me that Autodesk has developed some wonderful technology that is capable of doing some really wonderful things. But to see it packaged this way is quite disappointing.
In his review on the DEVELOP3D blog, Al Dean wrote:
What it boils down to is that because of the nature of whatever Autodesk is doing with Fusion and the edits it makes, there are at present, very explicit limitations in terms of what can and can’t be done when you’re moving data between Fusion and Inventor and hoping to have the history and feature tree reconciled and maintained.
This was after the Change Manager failed to come up with the correct solution after he asked it to accommodate four changes that he made in Inventor Fusion into the original feature tree. In a way Al is right. Of course there are limitations to what you can do with the technology. But unnecessarily over burdening it by making it accommodate all changes as once only goes to limit it even further, often to a point where it fails.
Whats ironic is the fact that users do not perform all direct modeling operations at once. But the poor Change Manager is being forced to perform a bunch of interlinked magic tricks simultaneously. I mean, in my experiment above, I first created the hole and then applied the chamfer. There was a time a time difference involved here, which could easily be used by the Change Manager to accommodate my individual direct modeling operations into the feature tree. As of now, this is not possible because of the way this technology has been packaged – two separate applications and two different file formats.
In the concluding part of this series I will compare Inventor Fusion with the Instant3D technology of SolidWorks.