alex-fung
commented
1 year ago Update: the assumption I had previously might be incorrect. I tried minting 10,000 NFTs onto a new tree again, but this time one at a time (took about 9 hours). I also had a loop checking the proofs every 30 minutes as these were minted – interestingly enough at a certain threshold it seemed that a bunch of the proofs got busted, wondering if it's a volume issue as opposed to a throughput issue in this case.
In particular, when a total of 7897 NFTs were minted, none of them had empty roots/proofs (although 12 did have invalid proofs), but by the time a total of 8663 NFTs were minted, 8193 of them were returned as empty. Curious if you have any thoughts regarding this. The tree for this latest experiment was 4A8wVYH2e3SPEsHUPTVcdrNGMmnZoPja9M7K9qxVqXyn
We're discovering that when minting 10,000 new NFTs to a tree at a rate of 10 per second, most the off chain proofs will be empty (highly reproducible).
Here's a script that demonstrates as such: https://gist.github.com/alex-fung/9f1963f34ab0ab9096653fd77965fc51 Note it takes about 20 minutes to fully run, but at the end when getting the proofs for each asset ~8000 of them are empty proofs. You will need to provide a keypair with enough SOL to create the 10,000 new NFTS (about 1.2 SOL), as well as an RPC endpoint and a ReadAPI endpoint