tpetrick
The issue will be fixed in the next daily build. In the mean time, you can work around it by accessing the task graph options through the regular Network Editor Display Options. If you press the ‘d’ key in the network editor, the dialog that comes up will have a Task Graph tab on it with the same preferences.

BrandonA
Currently only local machine services are supported, so it's not currently possible. However, adding support for PDG services to the HQueue Scheduler is on our roadmap.

tpetrick
I've attached an example file for you that shows all three cases. I've also logged an RFE for the Wedge node to automatically create a “wedgecount” attribute, since that's something we can provide internally.

tpetrick
That issue has been fixed in the latest build of H18.0. I've attached an example file with a ROP Fetch that cooks a file cache, and has the “Cook Frames as Single Work Item” option enabled. The single work item now properly lists cache files for each frame instead of just the first frame, and reports output files for each frame once the job completes.

This means that you can now use a Work Item Expand to create an item for each file. The work item expand doesn't need to be dynamic because the “Include Expected Outputs” toggle is enable on it.

tpetrick
If the ROP is producing other files on disk but not reporting them back to PDG while it cooks….

tpetrick
Are you able to attach an example file? What type of ROP are you cooking with the ROP Fetch node?

If the work item only lists a single file in the MMB menu, then that's the information PDG has to work with. If the ROP is producing other files on disk but not reporting them back to PDG while it cooks, then the only way to create work items from those files is with a File Pattern.

tpetrick
By expand, I assume you mean you want to create a work item for each output file in the ROP fetch work item? You can do that using a Work Item Expand node set to expand files, but it needs to know what the files are when it goes to generate. The output files on the work item are added while its cooking, so the Work Item Expand node will need to be set to Dynamic.

There's also an option to “Include Expected Output Files” on the expand node, which normally would let you use it statically. When “Cook Frames and Single Work Item” is enabled, however, the ROP Fetch doesn't add per-frame expected outputs. This is because that option is normally used when cooking a ROP that only creates a single output file, such as a USD Output, Alembic or FBX ROP. I can look into changing that so it correctly adds the per-frame expected output files to the work item, which would make it possible to use a Work Item Expand without making it dynamic.

Shadowjonny

tpetrick
You should just be able to cook it using the shift+g hotkey, or RMB -> Cook Node. Note that you'll need to change the Cache Mode parm to Write Files, otherwise the work items will be instantly marked as cooked because their output files are exist on disk.

I mean like when a attribute in wedge is changed it automatically recooks without me going in and tell it to make dirty and reCook.
Looking through PDG tuts and the Documentations my guess this is only through a python and calling the TOPs node and tell the graph to reCook?