aiken tx rerun with traces inplace

[waalge]

What is your idea? Provide a use case.

I'm running top quality dapp. Tracing is off, obvs.
Tx fails with plutus error. Eeeek. Can't diagnose - no tracing.
But i have very cleverly dumped the tx as cbor.
Now I ask aiken to run the problematic script again, but rather than run the original script, run a build of the script with tracing on.

Why is it a good idea?

Tracing where its needed.

What is the current alternative and why is it not good enough?

Crying / leaving tracing always on.

[MicroProofs]

I agree with this but wonder if it should be an offchain mechanism rather than through Aiken

[rvcas]

Yea honestly, might be better to remove aiken tx entirely

[waalge]

Fair

[KtorZ]

Orthogonal to whether aiken tx is still relevant, I like the idea of re-running with traces. It shouldn't fundamentally be too complicated, especially without any semantic change to the program. But I would even push the idea a bit further and simply allow re-running a transaction with a different validator? So one may even swap out an entire chunk of logic, and see if the results are different.

It becomes tricky with multiple validators referencing one another, but it also sounds doable for many scenario to just swap reference hashes on the fly in compiled PLC 🤔