Regulation.Bin Modding inconsistent

[ReinhardTristanHeydrich]

Hello.

O FromSoftware-RS Unfortunately, it does not currently implement the methods necessary to make parameter modification 100% consistent on any PC.

My PC has the following specifications:

CPU	Ryzen 3 3200g
GPU	RTX 3060 - 12 GBs
RAM	8 GB + 25 GBs Swap from SSD NVME
SSD NVME	2 Terabytes

So, both Elden Ring and Dark Souls 3 should be moddable using From-Software-RS, which is true, but not specifically for the function of modding game parameters.

So, what I had to do was implement a trait that implements methods to perform the correct Wait.

Since, when println this value in-game, even if using eldenring::util::system::wait_for_system_init, it doesn’t guarantees this fully loaded. Look this example of print I made modifying Is_fully_loaded to print the entries().count() and return false, and using my DLLs, which all which modifies the repository, uses wait_for_system_init before using my wait_until_loaded.

This also In the print and code, they are from my Elden Ring repository.

https://gitlab.com/ReinhardHeydrich/er-dlls

But the same rule applies to Dark Souls 3 also. In my memory, the correct number for verification in Dark Souls 3 was 45.

[ReinhardTristanHeydrich]

I was going to create a pull request for both the changes I made to From-Software RS to make it easier to modify regulation.bin, and also this one for wait_until_loaded, but since the situation regarding external contributions seems quite unclear to you all, i decided I will just create an issue.

[vswarte]

Hi

Feel free to PR whatever but listen to the feedback (as you'd expect). We have a bigger discussion going on about how to await the readyness of certain aspects of the statics however so I dont think we'll merge any such solution right now.

I would also really recommend using the task system to meter if your params are ready being loaded, this means you wont have to use any sleep on external threads (and you will not have data race issues). Counting the amount of params seems a bit suboptimal but I am not aware of a better way right now.

[nex3]

Is the parameter system ever initialized but in an incomplete during a running task? I'd be a little surprised if it were.

[vswarte]

Not sure actually now that you mention it

[vswarte]

Yeah so the params are loaded from a task itself it seems. CS::ParamStep which is derived from FD4TaskBase. Steppers are essentially just tasks with a statemachine taped on top.

                             **************************************************************
                             * CS::ParamStep::RTTI Base Class Array                       *
                             **************************************************************
                             CS::ParamStep::RTTI_Base_Class_Array            XREF[1]:     14333f5ac(*)  
       14333f5b8 f8 f5 33 03 20 f6       RTTIBase
                 33 03 98 f6 33 03 
                 08 f7 33 03 50 e5 
          14333f5b8 f8 f5 33 03     RTTIBase  CS::ParamStep::RTTI_Base_Class_D  [0]                                         XREF[1]:     14333f5ac(*)  
          14333f5bc 20 f6 33 03     RTTIBase  CS::CSStepTask<>::RTTI_Base_Clas  [1]
          14333f5c0 98 f6 33 03     RTTIBase  FD4::FD4StepTaskBase<>::RTTI_Bas  [2]
          14333f5c4 08 f7 33 03     RTTIBase  FD4::FD4StepTemplateBase<>::RTTI  [3]
          14333f5c8 50 e5 2a 03     RTTIBase  FD4::FD4StepTemplateInterface<>:  [4]
          14333f5cc c8 be 2a 03     RTTIBase  FD4::FD4TaskBase::RTTI_Base_Clas  [5]
          14333f5d0 a0 be 2a 03     RTTIBase  FD4::FD4ComponentBase::RTTI_Base  [6]

[nex3]

I wonder if that means that it does have an inconsistent state across tasks, because that state machine can be in multiple possible states (presumably as different file loads come in?). It might be worth doing an empirical test by starting a task that just logs how many params are loaded each time it runs.