RPC: add weight to mempool entry output

[esotericnonsense]

Tested against master using the REST api (/rest/mempool/contents), simple addition of a field.

Personal use case is for fee analysis software.

[promag]

No test affected 😞 care to improve by asserting the new field in the relevant RPC's 😉?

[laanwj]

Concept ACK

[MarcoFalke]

I guess you need to update the documentation as well.

[esotericnonsense]

Tests are failing.
txid2 and txid3 work as expected.
txid1 fails on both my test and #11203 (add wtxid to mempool entry output).

Do not merge as is.

[ajtowns]

txid1 is failing because at that point "tx" is actually referring to the input to txid1, not the transaction for txid1. It works fine for me if I add

tx = FromHex(CTransaction(), self.nodes[0].gettransaction(txid1)['hex'])

prior to the assert_equal lines (and uncomment them obviously).

[esotericnonsense]

Doh. You're absolutely right. Fixed.

The final commit 'Refactor segwit 3-tx-chain' changes all references to 'tx' to 'tx/tx1/tx2/tx3' in order to clarify that. It has a large diff and can be dropped if necessary (only affects code style).

I have also rebased on master at 3255d63.

Should be good to go now.

[morcos]

Concept ACK, but it turns out GetTxSize is not actually what we claim it is. See the calculation of GetVirtualTransactionSize which potentially includes number of sig ops in the calculation.

I thinke the right path forward is:

• a small BIP documenting this usage of virtual transaction size (which is used almost everywhere in our code
• documentation update pointing to this new BIP instead of 141 for defining virtual size
• this PR makes even more sense given this lack of easy conversion, but I think the newly added tests should at least be documented to note that they only work in the case that sig ops don't factor in, and perhaps we should add a test with a tx that has more sigops to show what happens in that case.

[luke-jr]

@morcos AFAIK that's exclusively used for node policy, and as such isn't a topic for standardisation...?