Onchain data fetching refactor #524
Conversation
gmbronco
commented
Nov 9, 2023
gmbronco
commented
- separating data fetching and parsing from database logic
- splitting into separate responsibilities (state, data, gyro)
- fixing protocol fee for pools in recovery - should be 0
- fixing BPT price rate
* simpler data fetching and parsing * splitting into separate responsibilities (state, data, gyro) * fixing protocol fee for pools in recovery - should be 0 * fixing BPT price rate
| multicaller.call(`${id}.totalSupply`, address, 'totalSupply'); | ||
| multicaller.call(`${id}.virtualSupply`, address, 'getVirtualSupply'); | ||
| multicaller.call(`${id}.actualSupply`, address, 'getActualSupply'); |
There was a problem hiding this comment.
I would move these to the pool type specific queries, otherwise its an extra 2 queries per pool that will either fail or are not needed.
There was a problem hiding this comment.
that's the part where we need to know the composable stable pool type, which depends on adding logic based on the factory address. i understand we don't need overfetching, but we fetch so much already that the cost is marginal. i tested both versions (this vs previous) and there is no perfomance impact.
| multicaller.call(`${id}.actualSupply`, address, 'getActualSupply'); | ||
| multicaller.call(`${id}.swapFee`, address, getSwapFeeFn(type)); | ||
| multicaller.call(`${id}.rate`, address, 'getRate'); | ||
| multicaller.call(`${id}.protocolYieldFeePercentageCache`, address, 'getProtocolFeePercentageCache', [2]); |
There was a problem hiding this comment.
Same here, but I dont see a way to have this without relying on the network context :)
|
|
||
| tokens.forEach(({ address: tokenAddress }, i) => { | ||
| multicaller.call(`${id}.tokenRate[${i}]`, address, 'getTokenRate', [tokenAddress]); | ||
| multicaller.call(`${id}.metaPriceRateCache[${i}]`, address, 'getPriceRateCache', [tokenAddress]); |
| tokens: result.poolTokens[0].map((t) => t.toLocaleLowerCase()), | ||
| balances: result.poolTokens[1].map((w, i) => formatUnits(w, decimalsLookup[result.poolTokens[0][i].toLocaleLowerCase()])), |
There was a problem hiding this comment.
can you change toLocaleLowerCase() to toLowerCase()? Just avoid any side effects
| weights: result.weights?.map(formatEther), | ||
| targets: result.targets?.map(String), | ||
| poolTokens: result.poolTokens ? { | ||
| tokens: result.poolTokens[0].map((t) => t.toLocaleLowerCase()), |
There was a problem hiding this comment.
I'm a fan of writing complete words for variables, we don't need to save any bytes :)
| } | ||
| if (multicallResult[pool.id] && multicallResult[pool.id].pausedState) { | ||
| isPaused = multicallResult[pool.id].pausedState.paused; | ||
| } | ||
| operations.push( | ||
| prisma.prismaPoolDynamicData.update({ | ||
| where: { id_chain: { id: pool.id, chain: networkContext.chain } }, |
There was a problem hiding this comment.
any chance we could also get rid of the network context while we're refactoring that?
There was a problem hiding this comment.
good point, moving it up to the pool service
