Tap context: restore d: as unit under MINIMALIF?

[landabaso]

Hey!

While working on Taproot support for bitcoinerlab and comparing other implementations, I noticed a typing difference.

In tapscript, MINIMALIF is consensus (BIP342), so d: can be unit.

I know you removed the u property from d globally in 2022 as the safe fix. Now that MINIMALIF is consensus in Tap, would you be open to restoring unit only for Tap?

I think this could be a small change: add a minimal_if() flag on ScriptContext (default false, Tap true) and a cast_dupif_minimal_if that sets corr.unit = true, used when minimal_if().

Drafted changes here: master...landabaso:rust-miniscript:tap-dupif-unit

Happy to open a PR if this is the right direction. Don't want to churn if the current behavior is intentional and I don't know your codebase very well.

[apoelstra]

I'm going to leave this issue open -- concept ACK adding u to d: in a Taproot context but I'd really like to avoid extending the ScriptContext trait right now.

Is this causing you script inefficiency right now? We can prioritize it if so. But if this is just about elegance/completeness I'd like to wait til we've done some refactorings of our script context.

[landabaso]

Is this causing you script inefficiency right now? We can prioritize it if so.

Not on my side. bitcoinerlab is a typescript miniscript/descriptors implementation and I was adding tapscript support and comparing outputs with rust‑miniscript. One fixture didn't match, which led me to dig into the 2022 decision and propose a small patch.

No urgency for me. It's mostly about parity/understanding. Happy to wait for the context refactor. Thanks for all the pioneering miniscript work!

[apoelstra]

Thanks for the poke -- I had thought I was waiting on review for the next PR in that refactor series, but actually I had forgotten to open it. So I opened the next one in #895 and we are moving forward again!