MAST is not actually related right here. Because the Ark vtxo timber are simply timber of sequential transactions. In concept they are often arbitrarily massive.
Some issues are to consider although:
-
Extra spherical individuals means extra interactivity. Since all individuals within the spherical have to supply signatures for the spherical to achieve success, extra individuals means greater chance a spherical will fail and must be restarted, so the longer a spherical will take, time-wise. Notice that ASPs can restrict the variety of individuals in a spherical if wanted and even run a number of rounds in parallel.
-
Extra vtxo tree leaves means the next price for unilateral exit. The variety of transactions a person has to broadcast (and pay charges for) for a unilateral exit is
log2(n)
withn
being the variety of leaves within the vtxo tree. Whereas this scales logarithmically, which is fairly nice, it might nonetheless change into a major burden for customers to take part. -
When broadcasting a number of transactions for a unilateral exit, most likely utilizing a price technique like CPFP (baby pays for guardian), mempool coverage places an efficient restrict on the dimensions of the exit transaction package deal. At the moment, Bitcoin Core has a descendent restrict of 25 transactions. Present work on cluster mempool will most likely change this limitation and change it with a cluster measurement restrict (IIRC of 100). The way in which both of those limits have an effect on Ark exit transactions is refined and relies on how many individuals are exiting from the identical Ark tx on the identical time.
Ultimately, I feel the latter consideration would be the real looking limiting issue on real looking Ark vtxo tree measurement.
For instance, utilizing pre-signed transactions (clArk, i.e. covenant-less Ark) or utilizing OP_CTV, this successfully limits the variety of nodes in your entire tree*. With a restrict of 25, that will imply a tree with as much as 24 leaves and with a cluster restrict of 100, that will imply 99 leaves.
Nonetheless, with extra highly effective introspection opcodes**, the intermediate leaves could possibly be fee-bumped in order that this restrict now not holds.
*: In actual fact if just one leaf of the tree can verify, it might make house for different branches. So for instance all nodes besides the leaves is the restrict. Which means that for n
leaves, there must be house for n + 1
txs within the mempool.
**: like OP_TXHASH, OP_CHECKTXHASHVERIFY, OP_TX, direct introspection opcodes (or CAT+CSFS)