Intercepting price updates #204

majabbour · 2022-07-22T18:50:28Z

Our design involves rersizing price accounts, as well as intercepting calls to instructions that update_the price in order to do the SMA tracking.

In this PR:

We intercept the add_price calls (but we do not change their behavior yet)
We add a no-op update_account_version instruction that is meant to update the version number of a given account, and resize price accounts.

The purpose of this is to get feedback on the structure of the program.

It also addressed two other issues:

added panic and heap definitions. (These are usually added in the entrpoint! macro is solana sdk, code won't compile without them once we start using anything that code panic like arrays, etc)
Updated the error message in the size of time machine assert so that it prints the correct size for better developer experience

majabbour · 2022-07-22T18:58:20Z

.

program/rust/src/lib.rs

program/rust/src/rust_oracle.rs

program/rust/src/lib.rs

guibescos

Let's use rust error handling and better import style.

* changed c return value of update price to indicate aggregation * forgot to add bindings * removed the need for casting * fixed update_no_fail * removed unnecessary docker packages

program/rust/src/lib.rs

program/rust/src/rust_oracle.rs

guibescos · 2022-07-26T20:19:20Z

program/rust/src/lib.rs

+    match unsafe { c_entrypoint(input) } {
+        0 => Ok(0), // Success
+        SUCCESSFULLY_UPDATED_AGGREGATE => Ok(SUCCESSFULLY_UPDATED_AGGREGATE),
+        2 => Err(ProgramError::InvalidArgument), //2 is ERROR_INVALID_ARGUMENT in solana_sdk.h


Can you use the version from the bindings instead of the magic number 2?

No, unless I make bindings of solana_sdk

* Make c robust to price account size (#201) * make pythd robust * made the oracle accept new size * left space for extra publishers * Make c robust to price account size (#207) * make pythd robust * made the oracle accept new size * left space for extra publishers * Intercepting price updates (#204) * changed c return value of update price to indicate aggregation (#190) * changed c return value of update price to indicate aggregation * forgot to add bindings * removed the need for casting * fixed update_no_fail * removed unnecessary docker packages * timeless coments * a convinient way to derive traits * changed importing style * added register_traits method * fixed formatting * add a new insctruction and interecepted update price calls * fixed formatting * more helpful error message for incorrect c account size * fixed merging errors * addressed some warning * updated comment * merging formatting issues * memory borrowing issues after rebasing * updated import * more import style changes * removed panics * bug introduced by merge * moved dispatching to processor * better formatting * cleaner error handling * error handling wrapper for c_entrypoint * removed unwrap * more comments * removed unused imports + changed import style Co-authored-by: majabbour <59592028+majabbour@users.noreply.github.com>

majabbour requested a review from guibescos July 22, 2022 19:06

majabbour force-pushed the intercepting-price-updates branch from fa92df1 to 548f1de Compare July 22, 2022 19:11