Use conditional weights | Substrate

Substrate provides a mechanism known as transaction weighting to quantify the resources consumed while executing a transaction. Typically, we use the weight functions returned from our benchmarking for this. But Substrate also allow us to apply a totally different weight function based on certain condition. We will walk through an example in this guide. Once defined, it can be used directly in your pallet, written as such:

#[pallet::weight(Conditional(\<your condition\>)

Objectives

Create and use custom weighting in your pallet.
Apply different weight functions based on certain condition on computing extrinsic's weight value.

Here are the different traits we'll be implementing:

`WeighData`: Weigh the data in a function. `pallet::weight` expects whatever implements `WeighData` to replace `T` with a tuple of the dispatch arguments.
`PaysFee`: Designate whether the dispatch pays a fee or not.
`ClassifyDispatch`: A way to tell the runtime about the type of dispatch being made.

Write the Weight struct

Open lib.rs file for your pallet in a text editor.
Import DispatchClass and Pays by declaring use frame_support::dispatch::{DispatchClass, Pays}.

Import weights primitives into the pallet.

use frame_support:: {
dispatch::{DispatchClass, Pays},
},
weights::Weight,

Declare a struct called Conditional and write an implementation of WeighData for Conditional where the first parameter is the condition that evaluates to a boolean value.

In the following example, if the condition is true, the weight will be linear to the input. Otherwise the weight will be a constant.

pub struct Conditional(u32);
impl WeighData<(&bool, &u32)> for Conditional {
  fn weigh_data(&self, (switch, val): (&bool, &u32)) -> Weight {
    
  // If the first parameter is true, then the weight is linear in the second parameter.
     if *switch {
        val.saturating_mul(self.0)
     }
  // Otherwise the weight is constant.
     else {
        self.0
     }
  }
}

Classify dispatch calls

Add dispatch::{ClassifyDispatch, DispatchClass, Pays} to your pallet's frame_support imports. Since this implementation requires a DispatchClass, use default to classify all calls as normal:

Open lib.rs file for your pallet in a text editor.

Import DispatchClass and Pays by declaring use frame_support::dispatch::{DispatchClass, Pays}.

use frame_support::dispatch::{ClassifyDispatch, DispatchClass, Pays};
// -- snip --

// Implement ClassifyDispatch
impl<T> ClassifyDispatch<T> for Conditional {
  fn classify_dispatch(&self, _: T) -> DispatchClass {
     Default::default()
  }
}

Implement Pays

Specify how Pays is used for the custom WeighData struct. Setting this to true require the caller of this dispatch to pay a fee:

Open lib.rs file for your pallet in a text editor.

Implement Pays for the Conditional structure.

impl Pays for Conditional {
  fn pays_fee(&self) -> bool {
     true
  }
}

Use the weighting struct for an extrinsic

Use the conditional weighting struct on your pallet's extrinsics like this:

#[pallet::weight(Conditional(200))]
   fn example(origin: OriginFor<T>, add_flag: bool, val: u32>) -> DispatchResult {
   //...
   }

Examples

Custom WeightForSetDummy struct in the Example pallet