Feature request: Checkpoint node disk caching

[WFab]

Since checkpoint node already has some “memory based” caching built into it now, maybe it can use the library format to cache data to disk too. It could be a separate “mode” altogether, though I could see it being implemented with very few parameters hidden behind “Advanced settings” checkbox.

Implementation:

1. It could be a simple checkbox “Cache to disk”. It would save a temporary “tml” dump to the “temp” folder set in world machine preferences, deleted on session close.

2. It could expose a “path” field, potentially opening up a possibility for “persistent” library saving, in case someone wants a world machine project to retain this disk cache across multiple sessions (for example, during a material building and testing run).

I personally am a fan of “clutter free” separate graphs. But if it’s a built in feature inside checkpoints, I guess more people will find a use for libraries. The code should already be there, may be simple to wrangle this one.

Thoughts?

[MP77]

+1 I would love to have this implemented as well. It would help me out immensely.

[hamzaaa1988]

+1 I think libraries are super cool but a bit tedious to use and something more automated would help tremendously - especially when running into RAM bottlenecks (which I like to do) :]

EDIT: Maybe this could not only be done for checkpoints but for groups or even single devices.

[Profwaldone]

+1

It seems that saved sessions are seposed to fill this role, but I have yet to see that actually work. Personally I would use the second version of your suggestion as I have graphs with multiple “biomes” that do interact with each other but almost never need to be fully rebuild from scratch which the current work flow does require me to do atleast once per workday.

[WFab]

@Profwaldone Have you tried “Library input” and “Library output” devices? You can use those to set up a custom checkpoint system. You can even bridge separate projects using a disk dump. I have a tutorial project explaining a part of this workflow HERE. Go through that conversation thread to find out the limitations of the current implementation.