Issue with conflicting FFI publishes

[EnriqueL8]

I was investigating the reason for the integration tests failing nightly and trying to fix them. I came across this E2E test https://github.com/hyperledger/firefly/blob/main/test/e2e/multiparty/ethereum_contracts.go#L331 which was constantly failing.

This test published a contract and waits for a series of events to occur. Unfortunately after looking at the logs of the nightly integration tests they timeout waiting for such events to arrive on the WS. I managed to replicate the issue locally by re-running that single test twice and extended the logs in the code to print the problem:

e2e.go:153: Message: &{{42a52273-cd85-4e04-801e-9287cda7a466  definition batch_pin {did:firefly:org/org_6042cf 0x19513597c0e423fc121bd275f2002fc78c7db27f} 2023-06-16T15:03:12.050928343Z default  10c825a38cbf050c85d7a19982b1755e1fe93936cf696be5d7b73c98b934a9ca ff_define_ffi e453008d4515cfb591297a934a7ea834c0ea2caa6fb3a5edf581ae09a6b7b606 <nil>} default 3d0a98c32bb68f5a759eb811713f64513675d7520424361bad7df0538e7f2fa3 08cc3db7-857f-434b-a662-c190605a1848 3e46daa9-6d26-4df1-945d-54b97101e73d rejected 2023-06-16T15:03:14.020196802Z FF10407: Rejected contract interface 'c0e03d4e-76f7-498c-ae75-97b1945a6ab2' - conflicts with existing: 1305816f-660a-40d3-9a8b-59c5ab465c68 [0x140002806c0]   0}

It seems that there is a conflict when we try and publish the FFI.

[awrichar]

After looking further into the logs, the rejection is due to an incorrect check in the code. The optimistic insert of c0e03d4e-76f7-498c-ae75-97b1945a6ab2 fails, and the code concludes this is because it conflicts with existing record 1305816f-660a-40d3-9a8b-59c5ab465c68.

This is not correct. Actually c0e03d4e-76f7-498c-ae75-97b1945a6ab2 is being published after being a previously-local-only FFI - meaning that the insert fails because of an ID conflict, which should take it down this branch to reconcile the published details with the local record:

firefly/internal/definitions/handler_contracts.go

Line 51 in 2fbd5b6

return dh.reconcilePublishedFFI(ctx, existing, ffi, isAuthor)

Item 1305816f-660a-40d3-9a8b-59c5ab465c68 is actually not a conflict at all. It has the same name but a different version. Seems to simply be an error in how we are interpreting insert errors.