Bound shares used for transactions

[adlerjohn]

The number of shares used for transactions, intermediate state roots, evidence, etc. should be bounded by some parameter. Otherwise, a malicious validator could create a maximum-size block filled with only transactions, and all full node would be required to fully download and process the block to compute the state.

[adlerjohn]

There is good reason to also separately bound the number of shares available to use for paying for messages, and all other requests with a reserved namespace ID: it avoids the two sharing a fee market.

[evan-forbes]

This was implemented when we added a max square size, and will be turned into a configurable parameter when we do #183

[evan-forbes]

I think I read this too quickly when closing, we still need to implement this

[rootulp]

it avoids the two sharing a fee market

If transactions have a different fee market then we may consider adopting a sparse share scheme for transaction shares (initially proposed in #1040). The rationale is that it isn't possible to determine exactly how many shares a PFB will occupy when using a compact share scheme. In order to meter a PFB for the number of shares they occupy, we would have to round up to the worst case number of shares. If transaction shares used a sparse share scheme, they could be metered based on the exact number of shares they occupy (similar to how blob shares are metered after #431).

Note: we could still have a separate fee market and meter transactions based on number of bytes rather than number of shares.

[adlerjohn]

Hmmm. Fee markets don't necessarily have to be played out entirely in-protocol. Ultimately, it's up to the block producers how they calculate priority.

[evan-forbes]

This can be closed now that the square is deterministic, which prevents filling the square with empty or nonsensical data, however there could still be a restriction put on how many txs could be included