[WIP] Feat staker v2 #63

baitcode · 2024-12-10T21:04:41Z

Added library for Unsigned FixedPoint calculations. UFixedPoint supports basic mathematical operations.

Added new method to Staker interface:

fn get_staked_seconds_at(self: @TContractState, staker: ContractAddress, timestamp: u64) -> u128;

Allows to calculate staked amount. If address staked 100 tokens for 10 seconds, that would be 1000 staked seconds. Fractions of a second are rounded down. Only full seconds count. Added additional data structure:

#[derive(Drop, Serde, starknet::Store)]
struct StakingLogRecord {
    timestamp: u64,
    total_staked: u128,
    cumulative_seconds_per_total_staked: UFixedPoint
}

and in storage

staking_log: Vec<StakingLogRecord>

and population logic into stake and withdraw methods. All searches in those logs are binary.

Closes: #61

…governor.

…support Q128.128 FP. Added cumulative seconds per total staked history value, and getter for that value Added method that allows to calculate how much staked seconds would staked token amount generate in between 2 timestamps: `calculate_staked_seconds_for_amount_between` Added tiny amount of tests

* removed staked seconds calculation * added seconds per total staked test and added logic for calculation of that value when nothing is staked

baitcode · 2024-12-14T22:55:20Z

@moodysalem Could you, please, review?

moodysalem · 2024-12-16T20:44:29Z

src/staker.cairo

+    #[derive(Drop, Serde, starknet::Store)]
+    struct StakingLogRecord {
+        timestamp: u64,
+        total_staked: u128,
+        cumulative_seconds_per_total_staked: UFixedPoint,
+    }


todo: store pack the timestamp and total staked into one felt252

My bad. Thank you for catching that. I should've used felt252 to store in a single slot.

moodysalem · 2024-12-16T20:45:10Z

src/utils/fp.cairo

+}
+
+pub(crate) impl FixedPointIntoU256 of Into<UFixedPoint, u256> {
+    fn into(self: UFixedPoint) -> u256 { self.value.try_into().unwrap() }


this really should be a try_into

Good point.

moodysalem · 2024-12-16T20:46:08Z

src/utils/fp.cairo

+    pub(crate) value: u512
+}
+
+pub impl UFixedPointStorePacking of StorePacking<UFixedPoint, u256> {


a u256 takes two slots, the fixed point number in this case is a 64.128 which is 192 bits which can fit in a single felt. would recommend not creating a type just for this unless it's specifically a FixedPoint64x128 type, but we haven't created special types elsewhere

Rookie mistake.

would recommend not creating a type just for this unless it's specifically a FixedPoint64x128 type, but we haven't created special types elsewhere

I though that having a small abstraction would not hurt, especially as it introduces zero cost. I understand that addition and subtraction are not needed and probably won't, but those seemed to easy to implement to omit those. Also having that as an abstract separate value allowed me for better testing. And having tests in place it was easy to refactor things. I just finished optimising the code, improving the storage and added special convenience method for creation of that value. This value is now FixedPoint124x128 (fits one felt252 exactly).

well. almost finished (fixing tests atm)

But you know. I'll try to make it a minimal size.

moodysalem · 2024-12-16T20:47:34Z

src/staker.cairo

+    #[derive(Drop, Serde, starknet::Store)]
+    struct StakingLogRecord {
+        timestamp: u64,
+        total_staked: u128,


probably better to store total_staked * seconds elapsed accumulator so we can get time weighted average total staked over a period. it makes the value ~32 bits wider but you can still always get the exact total staked amount at a particular time by using the time weighted average over 1 second

the reason is that the time weighted total staked is what counts as the total voting weight, not the total staked at a particular second

I finally got myself some piece of calm and though on that comment.

Short

I would suggest to store both, total_staked + weighed sum staked fields

Long

TWA - time weighted average.

We won't be able to calculate total_staked as TWA over 1 second.
While I was writing this I figured out that we still can calculate total_staked. But not for every case.

Let's imagine:

TS - timestamp
TWS - weighed sum staked, numenator of TWA
AMT - total staked
SPTS - seconds per total staked

Let's assume our stakes changes like that:

TS AMT * TWS SPTS

0 100 0 0

1 200 100 0.01

5 300 900 0.03

7 0 1500 0.03(6)

means that value is not stored in log and has to be calculated always

I'll use column name as a function that takes TS to search for row and returns value of column for that row TWS(0) = 0, TWS(5) = 900.

Let's imagine I want to calculate TWA at point 4:
As AMT is erased I'll need to calculate:

AMT(4) = (TWS(5) - TWS(1)) / (TS(5) - TS(1))

then I'll be able:

TWA(4) = (TWS(1) + AMT(4)*(4 - TS(1))) / 4

Ok. Good. This will work.

But if I need to calculate TWA(10) - the only way to figure out AMT(7) will be through SPTS.

I'll need to AMT(7) = (TS(7) - TS(5))/(SPTS(7) - SPTS(5))

And there will be a corner case for all the TSs < 0.

moodysalem · 2024-12-16T20:49:32Z

src/staker.cairo

@@ -104,6 +122,8 @@ pub mod Staker {
        amount_delegated: Map<ContractAddress, u128>,
        delegated_cumulative_num_snapshots: Map<ContractAddress, u64>,
        delegated_cumulative_snapshot: Map<ContractAddress, Map<u64, DelegatedSnapshot>>,
+
+        staking_log: Vec<StakingLogRecord>,


is there a limit to the stored vector length? we use maps elsewhere, e.g. delegated_cumulative_snapshot and delegated_cumulative_num_snapshots

I see. I don't think that size is the problem. But yeah seems that using Maps consumes less gas. Will do a bit of a research. Thank you for pointing that out!

I've done some research on that comment about representing Array of log records as a Map instead of Vec. Vec is actually similar to Map, with one distinction, it tracks count of elements. As we have append only log we don't need to worry about reindexing issues. Changing to Map will ruin the possibility to quickly find log record <= to the timestamp provided. The size restriction for Vec is 1 felt, which is 252 bits. I've calculated and even if we store data with 1 sec resolution we will have enough space to store data for 6.3*10^79 years. Which is actually ok. I think. I mean. Even if it'll become a problem I'll be dead by the time 8 ).

I've checked some more. While the documentation says that whole felt252 is reserved for Vector index, Vec api only allow u64 as index, which still allows for 573 * 10**9 years of data for second resolution.

moodysalem · 2024-12-16T20:52:40Z

src/staker.cairo

+            let total_staked = if is_add {
+                // overflow check
+                assert(last_record.total_staked + amount >= last_record.total_staked, 'BAD AMOUNT'); 
+                last_record.total_staked + amount
+            } else {
+                // underflow check
+                assert(last_record.total_staked >= amount, 'BAD AMOUNT'); 
+                last_record.total_staked - amount
+            };


if you only store cumulatives you can avoid this write in case where the timestamp is not changed

Makes sense.

Simplified while moving staker_log to separate file.

moodysalem · 2024-12-17T15:19:17Z

src/staker_test.cairo

+
+        set_block_timestamp(5000); // 5 seconds passed
+
+        assert(staker.get_staked(token_owner, delegatee) == 2, 'Something went wrong');


you should use the assert_eq! macro btw, no need to write error messages

I think that assertion messages lack line numbers, that is why I decided to stick with custom messages. But sure, no probs, will fix.

usually wouldn't matter because the assertion lines are never identical

you were right assert_eq! is much more convenient.

…ilerplate. Implemented custom packing and custom data SubPointers.

…ize. Added convienience method for u64 / u128 case.

finished subpointers for staking_log. Subpointer always read whole felt252 from memory and converts it. So I had to change approach for packing.

baitcode · 2025-01-09T22:23:49Z

@moodysalem Ok. I've made final changes. Looks like task is complete. Could you do the review, please?

moodysalem · 2025-01-20T11:39:30Z

#67 is new version

baitcode added 2 commits December 10, 2024 23:58

Initial implementation of staked seconds calculation

cb3e4dd

a bit of naming

01b3f73

baitcode marked this pull request as draft December 10, 2024 21:04

baitcode added 9 commits December 11, 2024 00:17

added upgrade method. NOTE: constructor changed. Staker now requires …

924aef3

…governor.

fix import

88f8db8

TODO: removal

9553035

remove todo

aedb403

test

9ee9324

local settings

c3c0244

revert upgrade code for now

692c36a

removed upgrade due to cyclic dependency with governor

349919e

finalise

d6504e9

baitcode marked this pull request as ready for review December 11, 2024 21:28

typo

5f32326

baitcode marked this pull request as draft December 11, 2024 21:46

baitcode added 3 commits December 13, 2024 21:55

removed old tests

2f9516a

* debugged fixed point math. added tests

bb4f99a

* removed staked seconds calculation * added seconds per total staked test and added logic for calculation of that value when nothing is staked

baitcode marked this pull request as ready for review December 14, 2024 22:54

moodysalem reviewed Dec 16, 2024

View reviewed changes

moodysalem reviewed Dec 17, 2024

View reviewed changes

baitcode added 8 commits January 4, 2025 02:16

some comment fixes

1a2d6fc

Extracted staker storage into separate file due to large amount of bo…

b18d702

…ilerplate. Implemented custom packing and custom data SubPointers.

rename staker_storage -> staker_log. moved all relevant logic there

7fdfb19

Simplified Fixed Point operations library. Reduced internal storage s…

b82953f

…ize. Added convienience method for u64 / u128 case.

simplified more. better naming

80cf041

clean up bitshifts from tests

f507631

overflow checks

164f64f

old tests did not respect overflow. had to create new ones.

6e931c6

finished subpointers for staking_log. Subpointer always read whole felt252 from memory and converts it. So I had to change approach for packing.

baitcode added 5 commits January 9, 2025 02:46

Debugging, cleaning up, fixing missed issues

fafba15

final fix

b55b320

final debugging

a28971c

cleanup

aa07801

cleanup

61f085c

baitcode added 2 commits January 10, 2025 13:21

removing UFixedPoint124x128 struct

f52fff1

BugFix for a case when fetl252 is overflown

4cc807e

moodysalem closed this Jan 20, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP] Feat staker v2 #63

[WIP] Feat staker v2 #63

baitcode commented Dec 10, 2024 •

edited

Loading

baitcode commented Dec 14, 2024

moodysalem Dec 16, 2024

baitcode Dec 17, 2024

moodysalem Dec 16, 2024

baitcode Dec 17, 2024

moodysalem Dec 16, 2024

baitcode Dec 17, 2024

baitcode Jan 6, 2025 •

edited

Loading

baitcode Jan 6, 2025

moodysalem Dec 16, 2024

moodysalem Dec 16, 2024

baitcode Jan 7, 2025

moodysalem Dec 16, 2024

baitcode Dec 17, 2024

baitcode Jan 6, 2025

baitcode Jan 9, 2025

moodysalem Dec 16, 2024

baitcode Dec 17, 2024 •

edited

Loading

baitcode Jan 9, 2025

moodysalem Dec 17, 2024

baitcode Jan 6, 2025

moodysalem Jan 8, 2025

baitcode Jan 9, 2025

baitcode commented Jan 9, 2025

moodysalem commented Jan 20, 2025


		set_block_timestamp(5000); // 5 seconds passed

		assert(staker.get_staked(token_owner, delegatee) == 2, 'Something went wrong');

[WIP] Feat staker v2 #63

[WIP] Feat staker v2 #63

Conversation

baitcode commented Dec 10, 2024 • edited Loading

baitcode commented Dec 14, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

baitcode Jan 6, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Short

Long

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

baitcode Dec 17, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

baitcode commented Jan 9, 2025

moodysalem commented Jan 20, 2025

baitcode commented Dec 10, 2024 •

edited

Loading

baitcode Jan 6, 2025 •

edited

Loading

baitcode Dec 17, 2024 •

edited

Loading