Core Lightning

global: more missing BOLT quotes.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

check-source-bolt: replace C check-bolt tool with Python check_quotes.py, check Rust and Python too

Extend check-source-bolt to also check Python and Rust source files.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Makefile: add check-requirements-coverage

Prints out a report of uncovered Requirements sections of the BOLTs.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

global: add the "easy" missing quotes.

Claude messed up about half of these: putting them too far from the
appropriate code. After this I did it myself.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

common/sphinx: don't create unused keys, fill in BOLT quotes.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

common: follow BOLT 4 requirements to make decryption constant time.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: use BOLT4's paranoid advice about doing constant-time error decryption.

lightningd: make sure we don't send channel_updates for unasked for channels.

We can decide to send an HTLC down a preferred channel which leads to the same peer
as the one they asked for, but the spec is clear that you shouldn't send the "wrong"
channel_update in that case.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: when we send errors, we won't include a `channel_update` if we chose a different channel than the one they told us to.

common: correctly refuse to accept wireaddr with port == 0.

BOLT 7 says this, but we don't do it. (Actually, it only says that for certain types,
but I've fixed that in https://github.com/lightning/bolts/pull/1331).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

gossipd: correclty ignore channel_announcement and channel_update for other chains.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: `gossipd` will now silently ignore gossip for other chains (rather than sending warnings).

channeld: update dust limit checks to the final version of PR 919.

This is what was merged in the spec, so update our checks (and bolt quote).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

patch fix-features-quote.patch

BOLT quote corrections in Python files (previously unchecked).

Update BOLT quotes in test and library Python files to match current BOLT
text: punctuation (periods to semicolons), capitalization (Lightning on
Bitcoin mainnet), fix field ordering in invoice breakdowns, add ... to skip
features fields and SHA256 hex details, correct backtick-quoting of field
names, and remove incorrect BOLT #7 label from a plain comment.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Fix C comments which will accidentally trigger Python BOLT quote checker.

We're going to get stricter, so "/* BOLT #N to-local output */" or "/* BOLT11 ... */"
will upset it.

Also remove a stray bare blank line in a BOLT comment block.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Update outdated BOLT quotes in channel open/close handling.

Some of these are malformed (thus were unchecked!), some are from the
stricter interpretation of `...` which won't cross section boundaries.

Several BOLT quotes had drifted from the current spec text:
- connectd/queries.c: BOLT #7 uses 'full_information' not 'complete';
remove a second quote that referenced query_channel_range but was
actually wrong (that requirement is for query_short_channel_ids).
- lightningd/dual_open_control.c: channel reserve is 1% rounded down
(not just 1%); witness weight check now says SHOULD broadcast rather
than MUST fail; RBF quotes simplified to match actual BOLT wording.
- openingd/openingd.c: 'The sending node' became 'The sender';
invalid signature response now says MUST send warning/error not just
MUST fail the channel.
- openingd/dualopend.c: same invalid-signature quote as openingd.c.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

check_quotes.py: add --coverage tracking; add devtools/bolt-coverage.py.

check_quotes.py gains --coverage=FILE: on each successful match, atomically
appends one line '{bolt} {section_idx} {start} {end}' to FILE using a single
os.write() call so parallel make invocations don't interleave records.

find_quote() and find_quote_immediate() are updated to return match start
positions (needed to record the covered range, not just the end).

bolt-coverage.py reads the coverage file and reports BOLT text not covered
by any source comment. By default it restricts output to Requirements
sections; --all-sections shows every section. --bolt N restricts to
a single BOLT number.

Exit status is 0 if everything in the selected sections is covered, 1 if
anything is uncovered.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

splice-script: Test for msat chan balances

Test splice scripts where the existing channel balance has fractions of a sat in them.

splice-script: Round channel balances down

When getting channel balances, we round off the msats for splice script.

Changelog-Fixed: Fix for `splicein`, `spliceout`, and `dev-splice` commands where channel balances included partial sats.

common/wireaddr: fix GCC 15 const-qualifier errors

GCC 15 propagates the const qualifier through strchr/strrchr, so the
result of strchr(const char *, c) is const char *. Declaring the locals
as char * triggers -Werror=discarded-qualifiers:

common/wireaddr.c: In function 'separate_address_and_port':
common/wireaddr.c:297:29: error: initialization discards 'const'
qualifier from pointer target type [-Werror=discarded-qualifiers]
common/wireaddr.c:304:27: error: assignment discards 'const'
qualifier from pointer target type [-Werror=discarded-qualifiers]

Make portcolon and end const char * to match; both are only read.

Fixes: https://github.com/ElementsProject/lightning/issues/9074
Changelog-Fixed: build: fix build errors with GCC 15 (Arch Linux).

doc: document UDS datagram tracing backend and envvar table

Add Backend 2 section for the new CLN_TRACE_SOCKET UDS datagram
transport. Restructure the existing USDT docs as Backend 1. Add
an environment variable reference table and a minimal Python
collector example.

test: Add a test for the UDS opentelemetry trace sink

trace: add UDS datagram socket backend, decouple from USDT

Add a new tracing backend that sends completed spans as atomic
datagrams over a Unix Domain Socket (AF_UNIX + SOCK_DGRAM), enabled
via the CLN_TRACE_SOCKET envvar.

Refactor trace.c so span management logic compiles unconditionally
rather than being gated behind #if HAVE_USDT. DTRACE_PROBE macros
are defined as no-ops when USDT is unavailable. When no backend is
active (no USDT, no trace file, no socket), all tracing functions
return early via disable_trace.

Changelog-Added: tracing: Add a unix-domain socket sink for opentelemetry traces

CI: add hook and notification schema paths to readme rpc sync

Changelog-None

pytest: fix flake in test_closing_anchorspend_htlc_tx_rbf

Fixes: #9088

```
FAILED tests/test_closing.py::test_closing_anchorspend_htlc_tx_rbf - assert (None or 3001.3083296990844 < (3000 + 1))
+ where None = did_short_sig(<fixtures.LightningNode object at 0x7f1a89bb9c90>)
```

The previous fix #9027 misdiagnosed the cause: the identical value
3001.3083296990844 appears with and without a short signature, so this
is not signature-length variation but deterministic floating-point
rounding in `fees / weight * 1000` for this specific transaction
structure. Replace the hand-rolled ±1 assertions with check_feerate(),
which already uses a ±2 window and has the did_short_sig escape hatch
built in.

Changelog-None

v26.04.1

lightningd: fix format specifier for bitcoin_tx_weight

Replaces %lu in channeld and spender splice calc_weight output so
format matches size_t;

Changelog-Fixed: Builds on linux/amd64 to push to Dockerhub