SendToRoute previously didn't accept invoice routing hints.
This was a limitation when paying a wallet invoice.
Invoice hints are now correctly taken into account.
We now accept incoming AMP payments that are paid by a combination of
htlcs and pay-to-open requests. All pay-to-open requests are combined in
a single request.
More symmetry would have been nice between legacy/trampoline (e.g. for
the `SendPayToOpen.origin` field) but it didn't seem possible.
Also added an expiry timestamp to prevent race conditions at the UI
level.
Writing the PaymentRelayed event to the DB is not atomic.
Each part is written independently of the others (one row per part).
This is fine as nothing relies on this event being written atomically.
However tests were expecting that and we would observe rare
test failures on travis.
When coming back online, re-send private channels' `channel_update`.
This makes sure it gets rebroadcast regularly in case it was missed.
Since it's a private channel, it won't spam the network.
We use the socks5 proxy that is defined in the configuration and is typically used to connect to LN nodes running as TOR hidden services.
This should allow users to connect to Electrum servers that are running behind TOR.
* Refactor payment errors
When sending payments, it makes it easier for a wallet to display
the correct localized error message to users.
* Faster Trampoline payments fulfill
We were previously waiting for the whole downstream payment
to be settled (all individual HTLCs).
We can do better and fulfill upstream as soon as we get the preimage
(which only needs one downstream fulfill).
We currently rely on `require`, which is convenient, but doesn't allow
fine-grained exception control.
Also, in case of errors, logging is done at the supervisor level, where
we lose the remote `node_id`.
Instead, we type some crypto-related errors and log them in the
`TransportHandler`, which already has the correct MDC.
The DB ordering is not deterministic.
For multi-part payments, the first timestamp is taken.
This can vary depending on which record is listed first.
Using the same timestamp avoids a failed assertion.
Instead of emitting this event when we send a signature, we emit it when
our `availableBalanceForSend` actually changes. This happens:
- when we send a new `update_*`;
- when we receive a `commit_sig`, which may acknowledge one or several
`update_*` sent by our peer.
We choose to only emit this event in `NORMAL` state, because its goal is
to give information about what payments we can make, which can only
happen in that state.
NB: other events `ChannelSignatureSent` and `ChannelSignatureReceived` give
a different type of information, and are sent in all states where
signatures are exchanges, not only in `NORMAL`.
The field `localBalance` has been removed because it was ambiguous, and so is
the balance tracking in the database, which wasn't very useful.
Co-Authored-By: Bastien Teinturier <31281497+t-bast@users.noreply.github.com>
* Electrum: allow watcher to watch for mempool transactions
Watcher now handles WatchConfirmed watches where min depth
is set to 0: the watch event will be sent when the tx enters the
mempool of the bitcoin node our Electrum server is connected to.
For 0-conf channel, use scids with a height of 0 and a tx index
generated from the first 16 bytes of the funding txid. This gives us
unique ids that can still be identified as 0-conf channel.
(cherry picked from commit 1734861930)
NB: this commit removes the phoenix special case for zero-confs channels
* Electrum: allow watcher to watch for mempool transactions
Watcher now handles WatchConfirmed watches where min depth
is set to 0: the watch event will be sent when the tx enters the
mempool of the bitcoin node our Electrum server is connected to.
For 0-conf channel, use scids with a height of 0 and a tx index
generated from the first 16 bytes of the funding txid. This gives us
unique ids that can still be identified as 0-conf channel.
Let a sender manually split a payment and specify a trampoline route.
Fix two flaky tests where the order of payment parts could be
different, resulting in a failed equality test.
If we're relaying multiple HTLCs for the same payment_hash,
we need to list all of those.
The previous code only handled that when Trampoline was used.
Comparing with the router ActorRef simply didn't work.
The reason is probably because Peers receive the router's supervisor ref
which doesn't match what `self` is inside `Router`.
Checking that the origin was the router felt brittle anyway.
We're now correctly typing the gossip origin.
We don't implement the upfront_shutdown_script feature.
However we update our encoding to always specify it.
This allows extending OpenChannel/AcceptChannel with tlv streams.
There is one caveat: Phoenix shipped with a version that's incompatible.
So we use a workaround to identify unpatched Phoenix versions
and send them the old encoding.
With MPP and Trampoline (and particularly the combination of the two),
we need to keep track of multiple amounts, recipients and fees.
There's a trampoline fee and a fee to reach the first trampoline node.
The trampoline nodes must appear in the route, but not as payment recipients.
Adding new fields to payment events and DB structs lets us distinguish those.
We also relax the spec requirement about feature graph dependencies.
The requirement to include `var_onion_optin` in invoice feature bits
was added after the first Phoenix release.
Phoenix users will thus have non spec-compliant invoices in their
payment history.
We accept invoices that don't set this field; this is a harmless
spec violation (as long as we set it in new invoices).
There was a rounding issue with the availableForSend/Receive calculation.
Because CommitTx fee and Htlc fee were computed separately,
but each was individually rounded down to Satoshis, we could
end up with an off-by-one error.
This resulted in an incapacity to send/receive the maximum amount available.
We now allow computing fees in msat, which removes rounding issues.
c-lightning fails to decode empty arrays of scids or timestamps with an encoding type set to COMPRESSED_ZLIB.
The spec is not specific enough on whether this is valid or not, so we'll set the encoding type of empty arrays to UNCOMPRESSED.
When paying an invoice, we weren't properly checking our own features.
If the invoice supported MPP, we would use it all the time.
If MPP isn't enabled in our features, we now default to a legacy payment.
(cherry picked from commit 60359c68e8)
When paying an invoice, we weren't properly checking our own features.
If the invoice supported MPP, we would use it all the time.
If MPP isn't enabled in our features, we now default to a legacy payment.
If our initial random deconnnection delay is 0 (unlikely but possible) then all "exponential backoff" reconnection delays will be 0 too, so we set a minimum value of 200 milliseconds.
(cherry picked from commit a0ae5ef13f)
