Any connected peer can send a `Subscribe` with `since` after `until`.
The service passes the range to `Store::filtered`, which asserts
`from <= to`. The panic stops the service thread.
Remove the assertion. The query already returns no rows when
`from > to`, so the peer gets no history.
There is no apparent reason why announcements with a timestamp of zero
cannot be stored. Avoid panicking.
Co-authored-by: Lorenz Leutgeb <lorenz.leutgeb@radicle.dev>
`disconnect_unresponsive_peers` drops any session whose `last_active` is
older than STALE_CONNECTION_TIMEOUT (hard-coded to 2 minutes).
`last_active` is written in only two places: `to_connected`, and
`handle_message` (the gossip message handler).
Pack data never reaches `handle_message` which is what marks . It arrives as `FrameData::Git`
and goes straight to the worker thread over a channel. So a peer busy
sending us a large pack appears completely idle, and
after two minutes we hang up on it mid-transfer. gitoxide then sees a
truncated stream and the fetch fails with:
Clone of rad:<rid> from <nid> failed with
'gix fetch error: Failed to consume the pack sent by the remote'
This only affects large repositories. Small ones finish inside the
window. Neither `limits.fetchPackReceive` nor `limits.fetchTimeout`
bounds it, and it reproduces against every peer, since any peer sending a
big pack goes gossip-quiet for the duration.
We now treat arriving pack bytes as liveness, symmetrically with the
`FrameData::Gossip` arm immediately above, which already calls
`received_message`.
Note that a remote that spends more than STALE_CONNECTION_TIMEOUT
server-side (e.g. enumerating or compressing a large pack before sending
its first byte), is still reaped, and `limits.fetchTimeout` does not help.
This is not a change in behaviour. Gossip silence would have reaped it anyway.
`VarInt` can get large enough to exhaust available memory on many
machines. Instead of allocating directly, check the size of the
allocation against some limit.
The limit chosen for frames is 128 KiB, which is roughly twice as much
as actual gossip and Git protocol frames are expected to be. However,
128 KiB is arguably short enough to not burden a standard consumer
machine too much.
This redundancy does not add much. These functions are already marked
via `#[test]`, and often times they are in a separate module named
"test" or "tests".
When `Service::handle_message` handles `Message::Subscribe`, constraints
regarding visibility of repositories are not considered, this leads to
leaking RIDs of private repositories.
Add a corresponding check.
Add a test to verify that reference announcements relayed via subscribe
correctly apply visibility filtering, there by ensuring references of
private repositories are only propagated to nodes on the allow list.
The test regarding deduplication needs a fix: Ensure that all mock
repositories are public and in storage so that they do not
get filtered out in case of unknown visiblity.
Co-authored-by: Lorenz Leutgeb <lorenz.leutgeb@radicle.dev>
Reasons for the rewrite:
- Migrate away from `ec25519` as our implementation of Ed25519,
because that crate is unmaintained.
Instead of `ec25519` we use `ed25519-dalek` which is the most popular
implementation of Ed25519 in Rust.
- Migrate away from `amplify`, because we make very little use of its
features. Here, we only use `amplify::Bytes32`, which is just a thin
wrapper of `[u8; 32]`.
Instead of `amplify::Bytes32` we use `type PublicKeyBytes = [u8; 32]`.
The realization that `ExtendedSignature` is not really tied to SSH, but
is merely a representation of a key together with a signature, allows to
relax the bounds on traits in `radicle-cob`.
Separating out uses of "public key" (a rather simple model of a key,
just 32 bytes) and "verifying key" (an actual point on the Edwards
curve, ready to verify signatures, using more memory than just 32 bytes)
allows a more sensible definition of `crypto::Signer`, which propagates
to `radicle-cob`, `radicle`, `radicle-cli`, `radicle-node` and
simplifies trait bounds.
In `radicle-node`, make the fact clearer that actually the secret key
itself is required (for our implementation of Diffie-Hellman) not
"just" signing. This results in removal of the generic argument
(usually called `G`) of multiple types.
Concretizing the types in `radicle-node` this way, i.e. by (morally)
setting `G = radicle_crypto::SigningKey`, leads to removal of
`MemorySigner` and `MockSigner`, as they really just wrap a signing
key.
Re-exporting `ed25519::Signature` (via re-export in `ed25519-dalek`)
is hopefully more future-proof than using our own type for signatures.
However, the fact that `ed25519::Signature` is not serializable and
does not implement `std::hash::Hash` means that we have to resort to
ad-hoc solutions in `radicle`:
1. The Identity COB requires serialization of signatures. These are
generally questionable, since the security of the Identity COB
should rely on the security of Signed References. In the medium
term we might even want to consider ignoring signatures in actions.
2. When reading Signed References, detection of Replay Attacks uses
a `HashMap` keyed by signatures. This detection only exists for
backwards compatibility.
These two use cases are exotic enough to warrant ad-hoc implementations.
By not wrapping `ed25519::Signature`, we encourage writing more widely
compatible code in the future.
Co-authored-by: Fintan Halpenny <fintan.halpenny@gmail.com>
This introduces a `fetch_timeout` setting to the node configuration,
replacing the hardcoded constant. It defines the duration a fetch stream
is allowed to stall before being aborted. The default remains 30s, but
users on anonymizing transports like Tor can now increase it to avoid
spurious timeouts during network congestion.
For most use-cases, i.e., except some in `radicle-node`, the channels
in `std::sync::mpsc` suffice, allowing us to get rid of the dependency
`crossbeam-channel`.
Crossbeam is sometimes mentioned to offer higher performance compared
to `std::sync::mpsc`.
For usage in `radicle-signals` (and the counterparts in `radicle-node`
and `radicle-term`) performance is irrelevant, as these channels are
used to send a singular message (to the effect of shutting down the
node or stopping the spinner). These could potentially use
`std::sync::oneshot` in the future, which is nightly-only as of
2026-06-17, see <https://github.com/rust-lang/rust/issues/143674>.
Usage in `radicle-protocol` and `radicle`, is a heavier duty.
The channels pass commands to the service (arriving e.g. via the
control socket in `radicle-node`), events from the service to other
listeners (e.g. the control socket), and fetch results from workers to
the service. For user interaction, performance is irrelevant, since
the messages are in the tens or hundreds per second. Even for delivering
events and programmatic usage of commands, thousands to hundreds of
thousands of messages per second are plenty.
Note that these channel implementations are benchmarked in the millions
of operations per second, see e.g.:
- <https://github.com/crossbeam-rs/crossbeam/tree/b23b7e8eca2efdad9bdc1ceb1aee1207a852c03b/crossbeam-channel/benchmarks#results>
- <https://github.com/fereidani/rust-channel-benchmarks/blob/b4c979a94482e5012734657c2ee153b528bc56d6/README.md>
Also, `crossbeam-channel` was merged into `std` in 2022-11, and we do
not believe that performance of the implementations has siginifcantly
drifted since then, see <https://github.com/rust-lang/rust/pull/93563>.
One minor downside is that the channels in the standard library do not
expose the number of pending messages.
The update to `flake.lock` is a simple
nix flake update rust-overlay
to be able to reach 1.97.1.
The changes to `**/*.rs` files are improvements suggested by Clippy.
The extension of the test shows that Node B's fetch cannot be dequeued because
the RID is in the active fetch set.
It can then be dequeued once Node A has been fetched from.
- Clear active[rid] only when the result's node matches the node that
started the fetch; mismatched results now report NotFound
- Prevent a late completion from a disconnected peer clearing a newer
fetch for the same repo started by a different node
- Add test covering the stale-from mismatch path
When a connection drops and the node attempts to reconnect within the
same second, the `last_attempt` timestamp written to the database is
identitcal to the `last_success` timestamp due to the Unix time (seconds)
precision loss.
The `maintain_connections` logic filters eligible peers using the
condition `last_success >= last_attempt`. Because the truncated
timestamps are equal, this evaluates to true, causing the node to
bypass the `CONNECTION_RETRY_DELTA` backoff.
If the OS instantly rejects the dial attempt (e.g. `EHOSTUNREACH`),
the node synchronously queues another connection attempt. This traps
the reactor in an infinite `while` loop.
Changing the condition to `last_success > last_attempt` ensures that a
failed attempt in the same second correctly evaluates to false, forcing
the node to respect the backoff timer and breaking the loop.
Add custom check for ellipses "...", asking for replacement of "…".
Git ranges and the CLI wildcard matches are ignored.
This change includes fixes to all sites that did not pass the check.
The invalid timestamp error will only tell the timestamp of the
offending node's message.
Currently, there are issues where that error is occurring but the
timestamp seems legitimate. This patch adds the running node's
timestamp to help debug the issue.
`cargo check` fails because `human-panic` and `sysinfo` require at least
1.88.0.
+++ command cargo check --release --locked --all-targets
error: rustc 1.85.0 is not supported by the following packages:
human-panic@2.0.6 requires rustc 1.88
sysinfo@0.37.2 requires rustc 1.88
Bump MSRV to fix this.
1.88.0 introduced [let chains], which in turn has `clippy` warn about
nested if statements. All of these sites are fixed in this change.
1.87.0 introduced [`is_multiple_of`], which is a more readable version
of `x % y == 0`.
[let chains]: https://blog.rust-lang.org/2025/06/26/Rust-1.88.0/#let-chains
[`is_multiple_of`]: https://doc.rust-lang.org/std/primitive.usize.html#method.is_multiple_of
Many usages of `Oid::from_sha1` are actually in the context of
generating arbitrary `Oid`s. In these cases, `impl Arbitrary for Oid`
may be used directly.
The length of a SHA-1 object identifier is defined in various locations.
Instead of this redundancy, make the definition in `radicle-oid` public, and
use it in `radicle{-{core,protocol},}`.
The main change that needs adjustment on our side is that
`impl EcPk for ec25519::PublicKey` changed from `Compressed = [u8; 32]`
to `Compressed = amplify::Bytes32`.
Also remove `impl Deref for PublicKey` to avoid dependents of
`radicle-crypto` depend on a particular implementation, and make the
anonymous member private.
See
<https://github.com/cyphernet-labs/cyphernet.rs/commit/f42396139a6321615dd651bc09073052add14a05>.
The main motivation behind this change is to get just a little
telemetry information from nodes on the network, namely the version
of Radicle they are running.
This is achieved by rewriting `impl Default for UserAgent` which now
uses the version information provided by the build script also used
in other crates.
Also, a new configuration option `node.userAgent` is added, which
allows users to override the user agent if they so please, or set
the value `null`, which will in turn send the user agent
`/radicle/`, which is not really helpful, and the default prior to
this commit.
Creations of `UserAgent` in the whole workspace is cleaned up. In
order to do that `UserAgent::test` is introduced.
The `paste` crate is no longer maintained, and the `pastey` fork is a
drop-in replacement.
`cargo deny check` output:
```
error[unmaintained]: paste - no longer maintained
┌─ /home/fintohaps/Developer/heartwood/Cargo.lock:239:1
│
239 │ paste 1.0.15 registry+https://github.com/rust-lang/crates.io-index
│ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ unmaintained advisory detected
│
├ ID: RUSTSEC-2024-0436
├ Advisory: https://rustsec.org/advisories/RUSTSEC-2024-0436
├ The creator of the crate `paste` has stated in the [`README.md`](https://github.com/dtolnay/paste/blob/master/README.md)
that this project is not longer maintained as well as archived the repository
## Possible Alternative(s)
- [`pastey`]: a fork of paste and is aimed to be a drop-in replacement with additional features for paste crate
- [`with_builtin_macros`]: crate providing a [superset of `paste`'s functionality including general `macro_rules!` eager expansions](https://docs.rs/with_builtin_macros/0.1.0/with_builtin_macros/macro.with_eager_expansions.html) and `concat!`/`concat_idents!` macros
[`pastey`]: https://crates.io/crates/pastey
[`with_builtin_macros`]: https://crates.io/crates/with_builtin_macros
├ Announcement: https://github.com/dtolnay/paste
├ Solution: No safe upgrade is available!
├ paste v1.0.15
└── (dev) radicle-protocol v0.6.0
└── radicle-node v0.18.0
└── (dev) radicle-cli v0.19.0
└── radicle-remote-helper v0.15.0
```
The intent to drop outgoing connections is modeled as `Option::None`
which is brittle and easy to miss.
Extend `enum AddressConfig` with a variant that is more explicit.
A timeout of PT3S was empirically found to be too short to allow for the
verification of Signed References, given that this verification is now
potentially much more costly compared to earlier versions.
PT30S should allow fetching reasonably sized histories, even if they
are on feature level "root".
Note that this is not a timeout for the overall fetch process, but
only for inactivity on the channel during an ongoing fetch.
When upgrading Signed References via `fn upgrad_sigrefs`, the value of
`RepositoryInfo::synced_at` was not updated. Thus, the caller could not
continue with migrated repositories as usual.
Now that enough information about Signed References is propagated, such
as the head of Signed References after upgrade, is available, it is
possible to construct `SyncedAt`.
Instead of repairing downgrades, automatic migration could fail.
This occurs when attempting to load signed references, and a downgrade
is detected while coalescing `RepositoryInfo`.
Similarly to how fetch recovers, automatic migration must also recover
by detecting for the downgrade error, and allowing the migration to
happen in this case.
Teach the `rad sync` and `rad clone` commands to accept the
`--signed-refs-feature-level` option. This option configures that
fetch to use the minimum feature level provided when fetching from
other nodes. This overrides the value of
`node.fetch.signedReferences.featureLevel.minimum`, and should only be
used in scenarios where it is necessary to downgrade the security of a
fetch for backwards compatibility.
To achieve this, the `Handle::fetch` method introduces an optional
parameter for threading through the `FeatureLevel`.
The configuration of the fetch process was only achieved at the setup
of the of the node.
This limits the ability to pass configuration options to the fetch
worker while the node service is running.
This change introduces `radicle_protocol::fetcher::state::FetchConfig`
to allow configuration options to be kept track of. This includes the
previous `timeout` option, but now also includes the
`radicle_fetch::Config`.
The service now produces the `FetchConfig` and passes it through to
the fetch worker.
In order to allow automatic migration to the feature level 'parent',
introduce a way to force writing signed references, even if the
user-controlled refs are unchanged.
To detect the special case of a Signed References history with a single
and root commit (which would otherwise be detected as feature level
'parent', even though the `refs/rad/sigrefs-parent` ref cannot be
written), keep information about the parent commit in `SignedRefs`.
The output of `rad inspect --sigrefs` is changed to match the more
strict interpretation of the feature level for histories with a single
and root commit.
To prevent from inability to write sigrefs in case of a verification
error, swallow verification errors about the head commit, log, and
carry on writing.
To be able to reason about backwards compatibility, downgrade attacks,
and future configurable enforcement of minimum required (security)
features, introduce the concept of a "feature level" in Signed
References.
In this implementation feature levels are assumed to be monotonic, i.e.,
that higher feature levels behave like lower ones, and only add
semantics on top.
Implementing this as an `enum` that is `Ord` flows nicely.
Treatment of the write-side is left for the next commit. Currently,
the write side "promises" `FeatureLevel::Parent`, but is not able
to "keep the promise", since it does not actually inspect the
feature level of the head in case it is unchanged.
CHANGELOG
The implementations for `Encode` and `Decode` for `SignedRefs` existed
for when `SignedRefs` was communicated over the wire. This was traded
for communicating `RefsAt` instead. So, these can safely be removed.