Fix underlyingBalance computing

[alexandre-abrioux]

Hi!

First of all, thank you for the dedication your putting on this project, the new protocol is awesome and well documented, it's a pleasure to work with.

I'm using aave-js and more specifically the pool-math helper to approximate users reserves balances from data previously fetched from the blockchain.

I found the library very useful to compute current stableBorrows with the getCompoundedStableBalance method, and current variableBorrows with the getCompoundedBalance method.

However I couldn't verify in my tests that the getCompoundedBalance method could be used to reliably compute the current underlyingBalance. With this in mind I tried to debug the method and to find a fix.

Test case

First of all I added a (failing) test case to try to verify the claim I'm bringing forward. You can checkout to the "add failing test case" commit to run it: cd01845

Fix

After that I basically translated to Javascript the logic found in the protocol:

• method calculateLinearInterest of MathUtils
• method getNormalizedIncome of ReserveLogic

; to create a new function currentATokenBalance specifically to compute the underlyingBalance.

You can find the new code in the commit "fix compute underlyingBalance": 3e2edb6.

Please give me your opinion on this, I may have overlooked something. Tell me what you think :)

Misc

I had to fix the eslint config to be able to commit because, as you can see here in the CI the command yarn lint is broken.

The issue comes from the fact that prettier/@typescript-eslint is still used in the project while it has been deprecated (see here).

However, I could not make it work by just modifying the .eslintrc.json, I think it is due to this issue: jaredpalmer/tsdx#498 (comment). tsdx doesn't seem to take into account the extends part of the .eslintrc.json config file. I had to migrate the config to the package.json file. It works for me now!

[sakulstra]

I think this is essentially an exact implementation of calculateCumulatedBalance from v1, let's ping @kyzia551 and ask what was the rational behind changing it in first place.

It's essentially the question, why the change from:
https://github.com/aave/aave-js/blob/master/src/v1/computations-and-formatting.ts#L207 to
https://github.com/aave/aave-js/blob/master/src/v2/computations-and-formatting.ts#L78

part of the changes seem to be the removal of redirection, but I have problems following the rational behind the other changes

[sakulstra]

@alexandre-abrioux thx for your efforts & investigating the issue! Good catch 🔍 👀

Is most of this functionality was already in v1 I created a pr to move it from v1 to common and reuse it in v2 #115.
The eslint changes are quite good as well (seems like these issues slipped in with merging all these eslint updates last week) - would you mind reopening a pr, cherry picking the lint changes?

[alexandre-abrioux]

@sakulstra Thanks for your answer! Nice find, I didn't work with v1 so I wasn't aware about that part of the SDK :)
Regarding eslint sure! I'll do this right away.
I'll close this PR then (PR moved to #115).