Implement a minimal fix for Penalty calculation in UndelegateClaimCircuit #3478
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cronokirby
had a problem deploying
to
smoke-test
GitHub Actions
Failure
cronokirby
force-pushed
the
min-staking-fix
branch
from
deff4f7 to
2083dcf
Compare
cronokirby
force-pushed
the
min-staking-fix
branch
from
2083dcf to
a696c62
Compare
cronokirby
had a problem deploying
to
smoke-test
GitHub Actions
Failure
cronokirby
force-pushed
the
min-staking-fix
branch
from
bb430cf to
5ffc155
Compare
cronokirby
force-pushed
the
min-staking-fix
branch
from
5ffc155 to
0b09542
Compare
|
Per discussion in today's sprint planning meeting, I'll add a commit later today that adds a test of the |
redshiftzero
temporarily deployed
to
smoke-test
GitHub Actions
Inactive
redshiftzero
approved these changes
Dec 7, 2023
| AmountVar { | ||
| amount: penalty_var.inner.clone(), | ||
| } | ||
| Ok(Self { |
There was a problem hiding this comment.
Here in AllocVar as an additional guard we could return an error if mode is not AllocationMode::Input
erwanor
reviewed
Dec 7, 2023
| use penumbra_num::{ | ||
| fixpoint::{U128x128, U128x128Var}, | ||
| Amount, AmountVar, | ||
| }; | ||
|
|
||
| /// Tracks slashing penalties applied to a validator in some epoch. | ||
| /// | ||
| /// The penalty is represented as a fixed-point integer in bps^2 (denominator 10^8). |
There was a problem hiding this comment.
We should replace this description
There was a problem hiding this comment.
We can remove all mention of how it is represented internally since the type is opaque
erwanor
reviewed
Dec 7, 2023
| .expect("converting integral U128xU128 into Amount will succeed") | ||
| } | ||
|
|
||
| /// Apply this `Penalty` to an some fracton. |
erwanor
approved these changes
Dec 7, 2023
There was a problem hiding this comment.
I have read through the PR and this looks correct to me. I saw some minor typos, I do think we could add more comments too: for example, during the review session, we spent some time recalling why round_down_to_amount would work over the upper limbs of the AmountVar, and shared some context about the endianness quirks of the underlying U256 impl. Otherwise, LGTM.
cronokirby
force-pushed
the
min-staking-fix
branch
2 times, most recently
from
5d529df to
3eaea83
Compare
hdevalence
approved these changes
Dec 11, 2023
This speeds up trusted param generation, because it was quicker to enable than to wait 16x longer
Now it is no longer possible to directly construct this type without going through a specific constructor which flags what a u64 actually means. Otherwise, surrounding code isn't changed. This also propagated to the circuit, which comes in at 15714 constraints, which is still under 2^14.
This saves on constraints: back to 14423 This also makes the circuit tests pass know, not sure why they were failing before.
This was a source of confusion about the representation.
Until we add a <= 1 constraint this is probably a good idea.
cronokirby
force-pushed
the
min-staking-fix
branch
from
554ab5f to
22ff965
Compare
cronokirby
temporarily deployed
to
smoke-test
GitHub Actions
Inactive
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Closes #3456.
I lost track of what exactly had changed in #3466, so I instead just implemented this without basing myself off of that PR.
The difference in approach is that this PR makes the
Penaltytype completely opaque, and modifies no other typed values in the system to use Penalty or U128x128 except the internal representation of the Penalty type.I think making Penalty opaque is the correct approach, because it makes it impossible for there to ever be any confusion about what the meaning of the u64s you're passing in to Penalty are.
Another difference with the other PR is that the circuit behavior should match the out of circuit behavior exactly now, whereas the other PR is missing a 1 - X operation in the circuit. This bumps up the constraint count to 15714. (We can golf this down with at least two strategies, but we don't really need to, it still fits).
One important thing that should be considered is the fact that this PR modifies the proto-encoding of penalty to now hold the full precision of U128x128 by encoding the 32 bytes that value consists of. I'm not quite sure of the ramifications of this change but it should be considered.
Update (2023-12-06T08:52:27+00:00): I've switched the Penalty to store the a fraction from [0, 1], representing the kept amount, rather than the removed amount. This simplifies compounding and the circuit, and also fixes the weird test failures I tried to debug for a couple hours.