Building Transaction with Assets to Mint

[emurgo]

Building Transaction with Assets to Mint

[leobel]

Hi, any update about that?

[flexabyte]

Hey thanks for the very prompt update! It turns out I was using the older version, I just started using the 10 beta and with a few changes to my TransactionBuilder setup I'm now able to mint using blockfrost. Tested pretty thoroughly with different policy scripts too and I'm fairly confident it's behaving as expected. Thanks for the great work!

[vsubhuman]

It's about to be merged and will be in the next version 10.0

See #231

[vsubhuman]

Hi, @flexabyte ! Ah yes, I think the burning part got forgotten when the processing of the Int values were added. Lemme add few changes for this and I will ping you once the new revision is available

[c4b4d4]

But if I do it like I stated on my first comment; add_change_if_needed() throws the following error:

missing input for some native asset

I was afraid of getting that non-balanced Transaction, and it indeed happened.

As far as I know, when minting multi-assets, you must specify in which output of the transaction you want to have this minted tokens. E.g, on the cardano-cli you do it as follows:

cardano-cli transaction build-raw \
       	--mary-era \
        	--fee 0 \
        	--tx-in dacb...7a5 \
          	--tx-out="+$LOVELACE+5 $POLICYID.couttscoin" \
        	--mint="5 $POLICYID.couttscoin" \
        	--out-file "$TX_BODY_FILE"

The transaction on this command gets balanced to 0 with the --mint argument, but in the case of cardano-serialization-lib as I stated on my first comment, the set_mint() comes after the TransactionBuilder.

Therefor, the calling of add_change_if_needed() throws missing input for some native asset because it does not know yet that it'll have Minted Tokens.

So as of today, it's not possible yet to use add_change_if_needed() when you going to Mint Tokens, and fees should be completely calculated manually with the given functions?

Would the transaction be built correctly even if the TransactionBuilderoutputs a non-balanced transaction?

[SebastienGllmt]

The TransactionBuilder doesn't support minting at the moment. You have to build the tx manually using the individual classes unless somebody adds support for it in the TransactionBuilder.

[Habacef]

would be interested too! (:

[flexabyte]

Hey @vsubhuman thanks so much for your work on this, I've been trying to do this in the background for a while and was amazed to see your branch. However, I've come across an issue and I'm not sure where it's coming from. I have written some JS that mints a simple token and submits it to blockfrost, and it works if I manually set the fee and change, but when using add_change_if_needed I get a fee too small error:

// THIS DOESN'T WORK
txBuilder.add_change_if_needed(
     S.Address.from_bech32(paymentAddr)
);
// THIS DOES WORK (after manually computing the outputs and fee)
txBuilder.add_output(
  S.TransactionOutput.new(
    S.Address.from_bech32(wallet.paymentAddr),
    S.Value.new(S.BigNum.from_str("15606829"))
  )
);
txBuilder.set_fee(S.BigNum.from_str("172453"));

The error:

ShelleyTxValidationError ShelleyBasedEraAlonzo (ApplyTxError [UtxowFailure (WrappedShelleyEraFailure (UtxoFailure (FeeTooSmallUTxO (Coin 172277) (Coin 172057))))])

I've not spent any time debugging this on the rust wasm side yet but just wanted to share anyway. I will update once I've looked into it further.

[vsubhuman]

Hi, @flexabyte ! Have you been using the latest version of the branch at the moment of getting this error? There previously was an error of not including the mint witnesses into the estimated tx size but now it were supposed to be fixed. Try it out plz on latest branch commit or on new published beta2: #231 (comment)

Also show plz if possible how are you creating the tx-builder (and the linear-fee parameter specifically) and how are you building the tx itself. The error (FeeTooSmallUTxO (Coin 172277) (Coin 172057)) suggests the calculation is 220 lovelaces short, which is 5 bytes in case you have fee coefficient 44.

Latest fix to the size estimation adds fake witnesses for each distinct policy ID in the mint: https://github.com/Emurgo/cardano-serialization-lib/pull/231/files#diff-67896f615f1e60f7474eb6194c0614a005522dc35fb07046855d77260954b972R99-R111

The tests show that adding these witnesses increases the size by:

No. of distinct Policy IDs * 32 + 3 bytes

Can see the test here: https://github.com/Emurgo/cardano-serialization-lib/pull/231/files#diff-67896f615f1e60f7474eb6194c0614a005522dc35fb07046855d77260954b972R3172-R3173

Quick look through the serialisation implementations for witness lists and hashes suggests that result makes sense, if there were no script keys apart from the mint (script witnesses for inputs are not supported yet anyways) - the witness set writes tag 1 to include the now non-empty set of script witnesses, native scripts then start the array and write the size of the array (the number of script witnesses) which must be the 2 additional overhead bytes, and the ScriptPubkey then writes an array of size 2 prefix (additioanal 2 bytes), adds an InsignedInteger tag 0 (which takes two bytes, if I am not mistaken) and writes the actual 28 byte hash of the pub-key, which gives the 32-byte per witness + 3 byte overhead.

Can't imagine so far where and why additional 5 bytes would be missing in that scheme. We will be doing some internal testing with the latest beta2 as well and will see if we are hitting a similar issue 🤔

[flexabyte]

Hey there, just for the sake of completeness, I've been trying to create burn transactions after successfully minting them and I've been having a few issues. Using the CLI you would normally build a transaction with a negative input, so I presumed I could use the add_mint_asset function that I use when minting, but just pass in a negative amount to be used as the input like so:

this.txBuilder.add_mint_asset(
            this.scriptHash, 
            S.AssetName.new(Buffer.from(assetName, 'hex')), 
            S.Int.new_i32(-1),
            S.Address.from_bech32(addr),
);

And I'm getting a mint.js:129 Uncaught (in promise) Asset amount cannot be negative!. When I dig into the debugger a bit I can see that the error arises when calling txBuilder.build(). The rest of my code remains unchanged from how it's used to Mint (i.e. I still call set_native_scripts on the TxWitnessSet etc.) so I'm not sure what could be the reason for it. Have you been able to successfully test burning by any chance?

Thanks!

[vsubhuman]

@flexabyte , I have made a couple of updates to that branch - can you plz check everything again now?

Btw, are you using it from Rust or do you need JS npm builds to try it out?