How does a Null object manage to have geometry associated with it when there appears to be no connections to it?

I was looking into fluids and floating things on them. In the AutoDopNetwork created after using ‘Embed in Fluid’ shelf tool there is a ‘Fetch Data’ node which somehow made the object float (that is what I was originally trying to figure out.) In the help page for that node there is the BuoyancyForce example which I launched. That is where I found the Mystery Null.

The Buoyancy Force example file's top level is a Digital Asset containing two things, a DOPnet (dopnet1) and a geo object (slug_source). In the dopnet there is a complex slug like geometry that is copied a few times in the form of RBD Objects whose SOP Paths (../../slug_source/SLUG) point to a single Null object. That Null is not connected to anything, and is locked. Unlocking it makes the geo in the RBDs disappear. It is this null which appears to pull a complex bit of geo out of nowhere that is the mystery.

The spreadsheet for the SLUG null shows a bunch of points, which I would assume are the vertices of the slug, but how did they get there? Copy input is checked on the SLUG null, but what input when nothing is plugged into it?

Obviously I am missing something basic in how these things are constructed but I have no idea where to look; help on Null objects has nothing that would suggest this is possible.

Any clues as to what is happening here? tanks…

locking the SOP node means caching the currently cooked geometry into the node itself, so it doesn't evaluate any inputs or parm changes anymore, the geo is simply locked

http://www.sidefx.com/docs/houdini14.0/model/lock_nodes [sidefx.com]

Thanks, that makes sense.

But what I do not see is how the data got into the Null Object to begin with. There are no input connections that are visible, and no tab's containing references to any geo.

Is it possible to lock the Null node and then delete the input connections, leaving the data cached inside a Null at the end of the chain?

answering my own question… yes:

“Locking nodes is memory efficient for most purposes, and has the convenience of allowing you to unlock the node later for editing. However, you can actually delete history for some node types, collapsing the effects of previous nodes into baked data, by pressing RMB on an object in the viewport and choosing Delete History.” - Houdini Docs

That must be what happened here. Looks like a handy trick.

yes, that's one way
but you can as well just lock node and then remove all the inputs, or copy it, paste somewhere else, plenty of ways how to get inputless node