feat: allow self payment

[shruti2522]

fix #2462

• implemented the following in outbound_payment.rs :

  • if the router tells us the payment is to ourselves, we build a simple one-hop route and treat it like a local payment
  • when preimage is available, immediately claim the HTLC

• added test case self_spontaneous_payment_fulfills() to test self payment

[ldk-reviews-bot]

👋 I see @TheBlueMatt was un-assigned.
If you'd like another reviewer assignment, please click here.

[TheBlueMatt]

This violates a number of the API constraints on sending payments - specifically, someone could send a payment, then restart and they wouldn't see the pending payment, thinking they hadn't "sent" it. Check out some of the early feedback on the first attempt at this #2573

[codecov]

Codecov Report

❌ Patch coverage is 92.55319% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 86.54%. Comparing base (7785360) to head (e5498e0).

Files with missing lines	Patch %	Lines
lightning/src/ln/outbound_payment.rs	92.55%	5 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4328      +/-   ##
==========================================
+ Coverage   86.53%   86.54%   +0.01%     
==========================================
  Files         158      158              
  Lines      103188   103280      +92     
  Branches   103188   103280      +92     
==========================================
+ Hits        89292    89385      +93     
  Misses      11471    11471              
+ Partials     2425     2424       -1     

Flag	Coverage Δ
fuzzing	37.03% <12.76%> (-0.05%)	⬇️
tests	85.84% <92.55%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[shruti2522]

This violates a number of the API constraints on sending payments - specifically, someone could send a payment, then restart and they wouldn't see the pending payment, thinking they hadn't "sent" it. Check out some of the early feedback on the first attempt at this #2573

I reworked this based on feedback in #2573 . We now try routing first and only fall back to a self-payment check if it fails, so explicit self-routes still work for rebalancing. I have also added phantom node support. this implementation reuses existing code without adding new ChannelManager fields. PaymentClaimable is still left since it’s a bigger change, should i implement it here?
please take a look @TheBlueMatt

[TheBlueMatt]

Yea, sadly I really don't think this is all that easy of a task - in order to provide the same semantics we provide for regular payments to self-payments (which imo is a strong requirement to land this) we need to have a disk write involved and keep the pending payment around like we do normal payments. Its not really entirely clear to me how we should do that - for a normal payment the way we do disk writes is via ChannelMonitorUpdates to the channel the payment is going out through. Of course for self-payments we don't intended to use any channel, so there's not a clear connected channel to use. We could of course pick some random channel to use, which is fine and maybe the right answer, but it does seem a bit awkward (I don't see a better answer tho tbh). Once we do that disk write we have to keep the payment around for a while until the events for the payment are processed and then remove it with another monitor update. On startup if we detect a self-payment that wasn't resolved we'll have to regenerate the events and replay them.

[shruti2522]

Yea, sadly I really don't think this is all that easy of a task - in order to provide the same semantics we provide for regular payments to self-payments (which imo is a strong requirement to land this) we need to have a disk write involved and keep the pending payment around like we do normal payments. Its not really entirely clear to me how we should do that - for a normal payment the way we do disk writes is via ChannelMonitorUpdates to the channel the payment is going out through. Of course for self-payments we don't intended to use any channel, so there's not a clear connected channel to use. We could of course pick some random channel to use, which is fine and maybe the right answer, but it does seem a bit awkward (I don't see a better answer tho tbh). Once we do that disk write we have to keep the payment around for a while until the events for the payment are processed and then remove it with another monitor update. On startup if we detect a self-payment that wasn't resolved we'll have to regenerate the events and replay them.

thanks for the explanation, I'm willing to implement the channel-based approach you explained, i just wanted to confirm should self payments only work when the node has at least one channel or should we handle the zero-channel case differently?

[TheBlueMatt]

Yea, I think its fine to just only allow self-payments if we have at least one channel. Devs shouldn't expect payments to work if they have no channels anyway.