Extra Crispy

[Mario Marengo]

Hi all,

I'm getting some strange double-cooks on one of my SOPs (or perhaps all of them...haven't checked the others yet).

This only happens when rendering, and when the following conditions are also true:

1. MySOP has the render flag set.

2. The display flag is set on MySOP or on any SOP further down the chain.

3. The input to MySOP is an HDA. (this isn't a condition but it makes the problem harder to detect, see below).

When the above is true, I get two consecutive cooks. On the first cook, isCookingRender() returns true (a Mantra cook) and checkChangedSourceFlags() reports that the source has changed when it hasn't (I've locked it). And on the second cook, isCookingRender() returns false (a Houdini cook), and again checkChangedSourceFlags() wrongly reports that the source has changed.

If the input to MySOP is changed to a regular SOP (as opposed to an HDA), checkChangedSourceFlags() reports correctly (that it hasn't changed), though I still get two requests for cooking (one from Mantra and one from Houdini)...

I want to avoid this redundant cooking since each one could potentially be expensive. I can hack my way around this with a "renderCook" counter, but I was wondering if there was a better detection method than checkChangedSourceFlags() (which fails with HDAs) to glean the real need for a cook.

Thanks.

[edward]

It's a bug that it's recooking when the subnet is locked. I assume that in your subnet, you have the display and render flags set on different nodes.

[Mario Marengo]

It's a bug that it's recooking when the subnet is locked. I assume that in your subnet, you have the display and render flags set on different nodes.

18804[/snapback]

DOH!

Yup, that was it: The subnet had render and display flags set to different nodes.

Thanks Ed!