Re-target Feature To New Object - Object Disappears

[Anonymous]

Hi there,

    I'll say straight off I'm used to SubD and procedural modeling (Modo, Houdini), and some MoI, so the Fusion 360 paradigm has, well, let's say it's not coming easily. It feels like trying to step backwards into layer-based work after experiencing node-based work - extremely limiting and lacking flexibility/power. Anyway.

The issue I'm running into is this: make a sketch, extrude it "to object" (i.e. to a construction plane). Do some other work, including making new planes. Then decide I need to change the target of the "to object" extrusion of that original sketch. However, when I go to do this (by right-clicking on the extrusion and saying "edit feature"), the new planes disappear...

I assume this is because I've "gone back" in the history before I added the new plane?? This seems... Insane, basically, from a design perspective. The two histories should be entirely separate!

Sure there is a way to do this...

[jeff_strater · Accepted solution]

"I assume this is because I've "gone back" in the history before I added the new plane??" - yes

"This seems... Insane, basically, from a design perspective. The two histories should be entirely separate!" - no

The reason why future features are rolled out is primarily to prevent you from creating circular dependencies.  This is a very complex check, so the most efficient way to prevent these cycles is to disallow "downstream" references.  You know those histories are independent because you created them.  It's not so easy for Fusion to know that, and to try to analyze this on the fly would create performance nightmares.  Every parametric CAD system works this way, and for the same reasons.

In your case, because that new plane is independent, you can drag it before the extrude in the timeline, and then be able to reference it when editing the extrude.

[Anonymous]

Aha, right. Thanks. So I can manually move independent features "back" in time. Annoying, but at least there's a workaround.

---

@jeff_strater wrote:

"This seems... Insane, basically, from a design perspective. The two histories should be entirely separate!" - no

The reason why future features are rolled out is primarily to prevent you from creating circular dependencies.  This is a very complex check

It's actually not complex at all - it only seems so because "history" is the wrong language to capture this kind of dependency information. Basically, the "history" is hiding a directed acyclic graph from the user. If you look at the steps in the history as nodes in a DAG, the problem is entirely trivial.

But programs which work with a DAG as the main interface modality exist (Nuke and Houdini being the two big ones), and there are no performance problems because it's trivial to ensure a graph is acyclic (i.e. has no circular dependencies).

It's too bad - there's no reason Fusion couldn't have been built with a DAG, it just wasn't, and it's the worse off for it (even if every other parametric CAD has the same limitation).

[Anonymous]

Anyway, thanks for the quick help. I don't mean to be a brat. Just frustrating finding something is a hassle when coming from a software package which does the same thing effortlessly!