Blobs Fee Adjustment #143
Blobs Fee Adjustment #143
Conversation
|
Current dependencies on/for this PR:
This stack of pull requests is managed by Graphite. |
|
|
||
| #[pallet::call_index(1)] | ||
| #[pallet::weight(0)] | ||
| pub fn update_parameter( |
There was a problem hiding this comment.
This call seems extremely dangerous and broad. What's the purpose of it?
There was a problem hiding this comment.
Yes, this call is undoubtedly dangerous. What I would like to achieve is simply a way to allow changes to the parameters of BlobAdjustmentFee with an extrinsic. Before using it, some necessary changes would be assigning proper weights and instead of accepting the key as bytes, it might be better to accept an enum that encapsulates both the key and value to avoid potential incorrect storage insertions.
The current implementation is just an experiment to demonstrate how parameters could be changed by governance/sudo
There was a problem hiding this comment.
There is already system_setStorage for setting storage arbitrarily - we should use that in favor of introducing another call.
There was a problem hiding this comment.
Uh, that's true! This makes that call completely useless. I will delete it
| // Define TargetBlockFullness based on our expectations | ||
| // of how we want our blocks to be full | ||
| // TODO: refactor | ||
| let avarage_blob_size = total_blob_size / total_blobs; |
| namespace_id: 0.into(), | ||
| blob: vec![], | ||
| }) | ||
| .size_hint() as u32; |
There was a problem hiding this comment.
this should be a constant
|
closing in favor of #149 |
This is an extremely experimental implementation, not ready for technical review but rather an example of a possible implementation.
The goal here was to test an approach to fitting
TargetedFeeAdjustmentto our needs.What I have done is create a connection between a
TargetBlockFullnessin bytes and aTargetBlockFullnessin ref_time.The main problem is that there is no easy connection between these two variables. One way I have figured out to deal with this is:
submit_blobextrinsic into two parts:base_weight(referring to all the extrinsic actions unrelated to the blob of bytes) andbyte_weight(the weight attached to each byte of the blob, which mainly depends on the hash function). The resulting formula for the weight is:base_weight + blob.len() * byte_weight.ref_timewill be composed, whether there will be more extrinsic with little blobs or fewer extrinsic with giant blobs.TargetBlockSizeif blocks were full more than expected.The final formulas are:
min_ext_len= size in bytes of an emptysubmit_blobextrinsicaverage_blob_size=TotalBlobSize / TotalBlobsaverage_submit_blob_weight= weight of asubmit_blobextrinsic with a blob the size ofaverage_blob_sizetarget_ref_time= (TargetBlockSize / (min_ext_len + average_blob_size)) * average_submit_blob_weightTargetBlockFullness=target_ref_time / max_available_ref_timeFinal thoughts:
I may not have considered some factors, and there may be different approaches that could be used rather than sticking solely to
TargetedFeeAdjustment.I am open to suggestions and criticism!