-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[KILO]: Refactor/evaluate htlc view #9097
base: master
Are you sure you want to change the base?
[KILO]: Refactor/evaluate htlc view #9097
Conversation
The purpose of this commit is to begin the process of packing symmetric fields into the newly introduced Dual structure. The reason for this is that the Dual structure has a handy indexing method where we can supply a ChannelParty and get back a value. This will cut down on the amount of branching code in the main lines of the codebase logic, making it easier to follow what is going on.
This commit begins the process of moving towards a more principled means of state tracking. We eliminate the mutateState argument from processAddEntry and processRemoveEntry and move the responsibility of mutating said state to the call-sites. The current call-sites of these functions still have their *own* mutateState argument which will be eliminated during upcoming commits. However, following the principle of micro-commits I opted to break these changes up to make review simpler.
This commit redoes the API and semantics of processFeeUpdate to make it consistent with the semantics of it's sister functions. This is part of an ongoing series of commits to remove mutateState arguments pervasively from the codebase. As with the previous commit this makes state mutation the caller's responsibility. This temporarily increases code duplication at the call-sites, but this will unlock other refactor opportunities.
In this commit we observe that the previous commit reduced the role of this function to a single assignment statement with numerous newly irrelevant parameters. This commit makes the choice of inlining it at the two call-sites within evaluateHTLCView and removing the funciton definition entirely. This also allows us to drop a huge portion of newly irrelevant test code.
This commit makes the observation that the nextHeight parameter of these two functions is no longer used by those funcitons themselves as a result of extracting the state mutation to the call-sites. This removes the parameter entirely.
In line with previous commits we are progressively removing the mutateState argument from this call stack for a more principled software design approach. NOTE FOR REVIEWERS: We take a naive approach to updating the tests here and simply take the functionality we are removing from evaluateHTLCView and run it directly after the function in the test suite. It's possible that we should instead remove this from the test suite altogether but I opted to take a more conservative approach with respect to reducing the scope of tests. If you have opinions here, please make them known.
This commit removes another raw boolean value and replaces it with a more clear type/name. This will also assist us when we later try and consolidate the logic of evaluateHTLCView into a single coherent computation.
This commit moves the collection of updates behind a Dual structure. This allows us in a later commit to index into it via a ChannelParty parameter which will simplify the loops in evaluateHTLCView.
This commit simplifies how we compute the commitment fee rate based off of the live updates. Prior to this commit we processed all of the FeeUpdate paymentDescriptors of both ChannelParty's. Now we only process the last FeeUpdate of the OpeningParty
We had four for-loops in evaluateHTLCView that were exact mirror images of each other. By making use of the new ChannelParty and Dual facilities introduced in prior commits, we consolidate these into two for-loops.
This further reduces loop complexity in evaluateHTLCView by using explicit filter steps rather than loop continue statements.
This commit observes that processAddEntry and processRemoveEntry are only invoked at a single call-site. Here we inline them at their call-sites, which will unlock further simplifications of the code that will allow us to remove pointer mutations in favor of explicit expression oriented programming. We also delete the tests associated with these functions, the overall functionality is implicitly tested by the TestEvaluateHTLCView tests.
Important Review skippedAuto reviews are limited to specific labels. Labels to auto review (1)
Please check the settings in the CodeRabbit UI or the You can disable this status message by setting the Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media? TipsChatThere are 3 ways to chat with CodeRabbit:
Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
CodeRabbit Configuration File (
|
Here we return the balance deltas from evaluateHTLCView rather than passing in references to variables that will be modified. It is a far cleaner and compositional approach which allows readers of this code to more effectively reason about the code without having to keep the whole codebase in their head.
1a3d207
to
2eb03bb
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice refactor, simplifying power of Dual
really shines 👌
lnwallet/channel.go
Outdated
h := update.addCommitHeights.GetForParty( | ||
whoseCommitChain, | ||
) | ||
|
||
return h == 0 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit: shorten?
return update.addCommitHeights.GetForParty(whoseCommitChain) == 0
lnwallet/channel.go
Outdated
uncommittedUpdates := lntypes.Dual[[]*paymentDescriptor]{ | ||
Local: fn.Filter(isUncommitted, view.OurUpdates), | ||
Remote: fn.Filter(isUncommitted, view.TheirUpdates), | ||
} | ||
|
||
return newView, uncommittedUpdates, nil |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
WDYT about making UncomittedUpdates(whose ChannelParty)
a method on HTLC view?
Saves the need for a closure + second return value here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Seems like an improvement. I'll tack on a commit to the end.
feeUpdates := fn.Filter(func(u *paymentDescriptor) bool { | ||
return u.EntryType == FeeUpdate | ||
}, openerUpdates) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A Last
op in fn
would be nice-to-have here: fn.Last(fn.Filter(...)).WhenSome(...)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Definitely. Thank you for giving me a nudge on it. I've been meaning to round out the API there.
if deltas.Remote >= 0 { | ||
theirBalance += lnwire.MilliSatoshi(deltas.Remote) | ||
} else { | ||
theirBalance -= lnwire.MilliSatoshi(-1 * deltas.Remote) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Not sure it would be better, but I'd be okay with a struct that expresses increase/decrease for readability:
type balanceDelta struct{
balanceIncrease lnwire.Millisatoshi
balanceDecrease lnwire.Millisatoshi
}
Could also have a getBalance(currentBal)
method that does the addition + removal so that it's a single line per-party here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In your imagined change here, would you allow both rows to be non-zero, or is this only a struct because go doesn't know what sum types are?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, allow both to be non-zero when we sum up the deltas and then add/subtract each accordingly.
Just seems easier to not deal with negative numbers when we're around lnwire.Millisatoshi
types.
Change Description
This is the next logical chunk extracted off the front of #8755. This is a pure refactor change. The core behavior of the code modified should be identical.
Steps to Test
make unit
Pull Request Checklist
Testing
Code Style and Documentation
[skip ci]
in the commit message for small changes.📝 Please see our Contribution Guidelines for further guidance.