use old ledger when determining target

[andymck]

During receipt v2 validations, the incorrect ledger was being used when determining the target. This could have resulted in valid receipts being rejected and is the probable cause for most of the "receipt not in order" errors.

thanks to community member BigEnigma for spotting this

NOTE: i have gated this behind the h3dex targeting chain var. The fix will come into use once h3dex is enabled

[BigEnigma]

Hey Mack

Looked at the change, not wanting to push my luck but one thing i'm not sure about shouldn't the targeting be based on the module defined in the vars/config at the time the transaction was run so its uses the right method?

BigE

[BigEnigma]

:) me again, sorry i noticed you applied this such that it would only kick in for target V6 and above, was this because prior to that we cannot reliable replay anything should anyone do so? this ones a curiosity question in just how to go about managing versioned logic against version transactions and data models :)

[madninja]

:) me again, sorry i noticed you applied this such that it would only kick in for target V6 and above, was this because prior to that we cannot reliable replay anything should anyone do so? this ones a curiosity question in just how to go about managing versioned logic against version transactions and data models :)

That's correct.. If we don't do this as part of v6 then replaying the chain would cause it to use newer code on older blocks which would be consensus breaking

[BigEnigma]

Hi

Did my first question about the variable used to get the targeting version make sense, is that bit ok going forward if we get to v7 but something gets an old txn from before a chain var switch should it not use v6 targeting instead of the current configured v7?

BigE