[XCM]: Remove fees mode

text/0106-xcm-remove-fees-mode.md

@@ -0,0 +1,78 @@
+# RFC-0106: Remove XCM fees mode
+
+|                 |                                                                                             |
+| --------------- | ------------------------------------------------------------------------------------------- |
+| **Start Date**  | 23 July 2024                                                                                |
+| **Description** | Remove the `SetFeesMode` instruction and `fees_mode` register from XCM                      |
+| **Authors**     | Francisco Aguirre                                                                           |
+
+## Summary
+
+The `SetFeesMode` instruction and the `fees_mode` register allow for the existence of JIT withdrawal.
+JIT withdrawal complicates the fee mechanism and leads to bugs and unexpected behaviour.
+The proposal is to remove said functionality.
+Another effort to simplify fee handling in XCM.
+
+## Motivation
+
+The JIT withdrawal mechanism creates bugs such as not being able to get fees when all assets are put into holding and none left in the origin location.
+This is a confusing behavior, since there are funds for fees, just not where the XCVM wants them.
+The XCVM should have only one entrypoint to fee payment, the holding register.
+That way there is also less surface for bugs.
+
+## Stakeholders
+
+- Runtime Users
+- Runtime Devs
+- Wallets
+- dApps
+
+## Explanation
+
+The `SetFeesMode` instruction will be removed.
+The `Fees Mode` register will be removed.
+
+## Drawbacks
+
+Users will have to make sure to put enough assets in `WithdrawAsset` when
+previously some things might have been charged directly from their accounts.
+This leads to a more predictable behaviour though so it will only be
+a drawback for the minority of users.
+
+## Testing, Security, and Privacy
+
+Implementations and benchmarking must change for most existing pallet calls
+that send XCMs to other locations.
+
+## Performance, Ergonomics, and Compatibility
+
+### Performance
+
+Performance will be improved since unnecessary checks will be avoided.
+
+### Ergonomics
+
+JIT withdrawal was a way of side-stepping the regular flow of XCM programs.
+By removing it, the spec is simplified but now old use-cases have to work with
+the original intended behaviour, which may result in more implementation work.
+
+Ergonomics for users will undoubtedly improve since the system is more predictable.
+
+### Compatibility
+
+Existing programs in the ecosystem will break.
+The instruction should be deprecated as soon as this RFC is approved
+(but still fully supported), then removed in a subsequent XCM version
+(probably deprecate in v5, remove in v6).
+
+## Prior Art and References
+
+The previous RFC PR on the xcm-format repo, before XCM RFCs were moved to fellowship RFCs: https://github.com/polkadot-fellows/xcm-format/pull/57.
+
+## Unresolved Questions
+
+None.
+
+## Future Directions and Related Material
+
+The [new generic fees mechanism](https://github.com/polkadot-fellows/RFCs/pull/105) is related to this proposal and further stimulates it as the JIT withdraw fees mechanism will become useless anyway.