I have a LOPs HDA that is a bundled preview renderer for assets in our pipeline. It sorts prims, assigns some utility materials, and then renders + composites images with a watermark via an internal TOPs network.

I gave the node two modes it can run in: "In-Process" and "Out-of-Process", so that artists can render cheaply in their current Houdini session and save some time vs. batch out bigger jobs to separate hython processes.

Something odd happens when cooking with "In-Process" selected: after the TOPs have completed cooking, the main HDA unlocks itself. This isn't great because this HDA is under active development and will wind up in dozens of scenes. The artist can always "Match Current Definition" to get up to date with any version changes but it's one more thing that could go wrong.

I thought maybe designating the internal topnet as editable in the HDA would help but it hasn't affected the unlocking behavior, and I don't really want to leave it editable anyway.

Any idea why this happens? Or at least what part of the puzzle I should be looking into to solve it? Random guesses: I'm using Python processors in the topnet to manage work item variables; also using Python Script LOPs in the HDA to parse info about the incoming scene and set context options that are used in the render settings + TOPs.

I've also noticed that the usdrender_rop LOP node that is called by the ropfetch always has it's frame range reset to the defaults after every run. The topnet/ropfetch is controlling the frame range so it doesn't matter (outputs are correct) but it's odd that it's changing a parameter on the LOP node.