Refactor interfaces

[QGarchery]

Refactor interfaces with the following goals in mind:

• provide a single interface (IMetaMorphoFull) that enables you to do all the calls to MM
• provide an interface (IMetaMorpho) that the MetaMorpho contract inherits so that we benefit from the compiler checks on it
• factorize the two interfaces so that we don't write the function signatures twice
• fixes [protocol review] Provide an interface to get all structs #216

[MerlinEgalite]

But then IMetaMorpho does not inherit IERC4626 and thus the documentation won't appear in etherscan because the following comments won't work: /// @inheritdoc IERC4626

[QGarchery]

I think it works because MetaMorpho inherits from ERC4626 which itself inherits from IERC4626 (btw this is one other advantage of this refactor: function signatures are imported only once, which is clearer). In this test PR, you can see that if you remove is ICounterGetNumber you get an error in the inheritdoc comment, which means that the inheritdoc comments are aware of inherited interfaces of inherited contracts

[MerlinEgalite]

I'm not sure convince yet this is a better way to split the interfaces per the comment raised above

[QGarchery]

I'm not sure convince yet this is a better way to split the interfaces per the comment raised above

It could be refactored a bit (comments, naming, ...) but I think that having a single interface for every usage is very nice, I don't see any real downside for now. Maybe we should even use the main name IMetaMorpho for the biggest interface with all the functions (which is called IMetaMorphoFull for now). Indeed that's the one that is expected to be useful for integrators, while the one inherited by the contract MetaMorpho is mainly there for checking the function signatures against the implementation

[Rubilmax]

I'm not sure convince yet this is a better way to split the interfaces per the comment raised above

It could be refactored a bit (comments, naming, ...) but I think that having a single interface for every usage is very nice, I don't see any real downside for now. Maybe we should even use the main name IMetaMorpho for the biggest interface with all the functions (which is called IMetaMorphoFull for now). Indeed that's the one that is expected to be useful for integrators, while the one inherited by the contract MetaMorpho is mainly there for checking the function signatures against the implementation

That is a great idea! IMetaMorpho -> IMetaMorphoFlat or IMetaMorphoStaticTyping and IMetaMorphoFull -> IMetaMorpho

we should do the same for Blue IMO

[QGarchery]

Changed all the names, please review and tell me what you think:

• IMetaMorpho is now the full interface, expected to be the most used
• IMetaMorphoStaticTyping is the interface inherited by the MM contract (and used only there). It is checked by the compiler against the contract
• IMetaMorphoBare the smallest interface that is common to the 2 previous interfaces (and used only there)

Should we put those in separate files ?

[MerlinEgalite]

It's better now. I like @Rubilmax naming suggestion with base.

Perhaps we should add comments in interfaces to guide integrators? It may be hard to know which one to use at first glance

[MerlinEgalite]

@QGarchery will you apply the change?

[MerlinEgalite]

The CI is broken can you update this PR please @QGarchery ?

[QGarchery]

The CI is broken can you update this PR please @QGarchery ?

I merged the target branch but with no luck. Not sure what the problem is, why can't it find morpho-blue-irm ?

[MerlinEgalite]

Ok I get it this is because of the IRM dependency. Just created this issue #266 and assigned @Rubilmax to it

[MerlinEgalite]

Please do not merge pending PRs on cantina-review

[MerlinEgalite]

Reverting the PR here #271

[QGarchery]

Re-open the PR in #272