- Drop `type='ipv6'` rows whose bracket-stripped inner part contains
no `:`, since every valid IPv6 textual form has at least one
- Cover the offender (`[]:8776`), bracketed garbage (`[abc]:8776`),
the unspecified (`::`), loopback, and full forms in tests
- Keep `dns` and `ipv4` rows untouched
Migration 8 retyped any `[..]:N` dns row to ipv6 by bracket pattern
alone, without checking the inner part is parseable
- Decode each row independently so one corrupt entry no longer aborts
the whole address-book iterator
- Log a warn with the raw `value` and continue past parse failures
- Add a regression test that mimics the post-migration-8 state where a
legacy `[]:8776` row lands as `type='ipv6'`
A pre-`df8e4e6c` parser admitted `[]:8776` as `HostName::Dns("[]")`.
Migration 8 retyped any `[..]:N` dns row to `ipv6` without validating
the inner part, so on read `Address::from_str` rejects the row with
"invalid IPv6 address syntax", which bubbled up and caused
`available_peers()` to silently return empty, breaking peer discovery
for the affected node.
Update the output of `rad_jj_colocated_patch` so that it passes with the current version of `jj` being used:
```
jj --version
jj 0.35.0
```
Mark `rad_jj_bare` as `ignore` since it currently cannot determine the
RID from the `rad` remote.
The name `print_inline` was used because, previously, the `print`
function name was taken.
This has now changed, allowing the rename of this function to happen.
The name `print` was used because, previously, the `println`
function name was taken.
This has now changed, allowing the rename of this function to happen.
Prevent `rad` from panicking when its stdout is a closed pipe, e.g.
when running `rad config | head`.
The broken pipes are handled at the output boundaries using three
layers of defence:
1. Inner: Route all stdout output through `term::print()` and
`term::print_inline()`, which use explicit `writeln!`/`write!` to
a locked stdout and silently ignore write errors. All `println!`
calls in `radicle-cli` and `radicle-term` are converted.
2. Middle: Catch `BrokenPipe` errors that propagate up via `anyhow`
from subcommands in `main::run()`, and exit cleanly with code 0.
3. Outer: Install a panic hook that intercepts `println!` panics
containing "Broken pipe" (from dependencies like clap) and exits
cleanly, chained in front of human_panic's hook.
To prevent regressions, `#![deny(clippy::print_stdout)]` is added to
`radicle-cli`, requiring all future stdout output to go through the
safe `term::` functions.
Note: `eprintln!` calls (stderr) are not handled, as broken stderr
pipes are extremely rare in practice.
Rust (since 1.62) ignores EPIPE by default (see [Rust #62569]),
causing writes to closed pipes to return `io::ErrorKind::BrokenPipe`
errors. The `println!` macro panics on these errors, producing a
confusing backtrace instead of the silent exit expected of Unix CLI
tools.
To avoid the use of `print!` and `println!`, the `radicle-term`
helper functions use `write!` and `writeln!`. This lays the groundwork
for the `rad` CLI to transition to using only `radicle-term`
functions.
There was a first attempt made, which is documented here. A
process-wide SIGPIPE reset (`SIG_DFL`) was ruled out because `rad`
manages child processes via libgit2 pipes and communicates with the
radicle node over Unix sockets. Resetting SIGPIPE globally caused
`rad` itself to be killed during internal pipe operations (e.g. during
`rad remote add --fetch`), producing flaky test failures.
[Rust #62569] https://github.com/rust-lang/rust/issues/62569
Add tests that verify the `rad` binary does not panic when its stdout
is a broken pipe (e.g., `rad config | head -1`).
Rust (since 1.62) ignores SIGPIPE by default (see [Rust #62569]),
causing `println!` to panic with "failed printing to stdout: Broken
pipe" instead of exiting silently. This is the standard Unix behaviour
expected of CLI tools.
These tests currently fail on `rad config` and `rad self`, and are set to `ignore`, to be enabled when fixed.
[Rust #62569]: https://github.com/rust-lang/rust/issues/62569
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},}`.
A small refactor to move this constant closer to the `enum` that it is
about. An associated constant constant fits more nicely.
Reason for changing the name from `SHA1_*_LEN` to `LEN_SHA1`: As we
think about the object format in the future becoming variable, we
would have that the length of an OID is parameterized by the object
format. So, we imagine a function `fn len(format: ObjectFormat)`, to
be called as `Oid::len(ObjectFormat::Sha1)`. This is the new order
we use.
When the underlying hash function is not important, but only the fact
whether the OID is the zero sentinel, use `Oid::is_zero`. This is more
flexible, as it is forward compatible with the SHA-256 object format.
`fn sha1_zero` can be `const fn`, and since it does not take arguments,
it can just be a `const`. There is no good reason to not have a constant.
Justification for changing the order to first specify "zero" and then
"SHA1": In the future we will support multiple object formats. A
function that produces a zero OID would be parameterized by the object
format then. Imagine `fn zero(format: ObjectFormat)` which would be
called as `Oid::zero(ObjectForma::Sha1)`. In anticipation, we change the
ordering to match.
Because we changed our main domain from radicle.xyz to radicle.dev, we
should also change our systemd credential identifiers.
Handling is move into a new function, which also checks the old names
for backwards compatibility.
With `connect: std::collections::HashSet`, we suffer unpredictable reordering
of values in our test cases.
Use `connect: indexmap::IndexSet` instead, which preserves insertion
order for iteration.
The `.envrc` file acts as a personal setup file, and should not be checked in.
The team provides `.envrc.sample` where contributors can choose to:
echo 'source_env .envrc.sample' >> .envrc
To bootstrap their own `.envrc` file.
Fixes an intermittent failure caused by a race condition with background
gossip messages.
Previously Eve's `RefsAnnounced` message could be processed by Bob
after the test updated the identity document to include Alice-Laptop
as a delegate. This caused Bob to unexpectedly fetch the issue in the
background, failing the `!has_issue` assertion.
We now explicitly wait for Bob to receive Eve's announcement before
updating the identity document.
Avoid relying on a thread sleep to wait for the nodes to connect.
Instead rely on the connection condition being established by using a polling loop.
The loop is controlled by a number of iterations, panicking if the
number of iterations is exceeded.
Addresses intermittent failures in the `test_connection_crossing` e2e test,
which were particularly prevalent on slower CI environments (such as
`rust:trixie`).
Previously, the test spawned two threads to make Alice and Bob dial each
other concurrently and strictly asserted that the "preferred" peer (the
one with the higher Node ID) would always win the `Outbound` link direction.
However, this assumption is flawed in real-world, OS-level network execution
due to two race conditions:
1. Thread Scheduling: One thread could execute and fully establish a
connection before the other thread even began processing its dial command.
2. Reactor Event Ordering: A node's reactor might wake up and process an
incoming TCP connection from its peer *before* it processes the `Connect`
command sent by the test. When it finally processes the `Connect` command,
it sees a session already exists and skips dialing entirely.
In both scenarios, a true "simultaneous crossing" never occurs. Instead, a
standard sequential connection happens, meaning the link direction is dictated
by whoever dialed first, not by the "preferred" peer logic. This caused the
strict `left: Outbound, right: Inbound` assertions to panic.
To fix this, two changes were made:
- Introduced a `std::sync::Barrier` to synchronize the two test threads.
This forces both threads to wait for each other before calling `.connect()`,
maximizing the probability of a true simultaneous dial.
- Relaxed the final assertions. Because OS-level TCP handshakes and reactor
polling can never guarantee perfect simultaneity, we no longer assert
*which* peer gets the `Outbound` link. Instead, we assert the core invariant:
that exactly one connection is established between the nodes, and that their
link directions are opposite (`s1.link != s2.link`).
Two tests assumed that `refs/remotes/rad/HEAD` would automatically be
crated. It turns out, however, that this only the case if executed
with Git 2.48.0 or newer. Git versions older than 2.48.0 do not touch
'refs/remotes/<remote>/HEAD'.
Git 2.48.0 and newer will update `refs/remotes/<remote>/HEAD` by
default. This can be disabled by setting
`remote.<remote>.followRemoteHEAD` to "never". Thus, with these
versions, setting that configuration to "never" emulates the behaviour
of older versions.
Therefore, to ensure that test results are consistent across Git
versions before and after 2.48.0, the fixture of test repositories
now include setting 'remote.rad.followRemoteHEAD = never'.
The affected tests are adjusted accordingly.
This change can possibly be reverted once Radicle requires usage of Git
2.48.0 or newer.
Co-authored-by: Lorenz Leutgeb <lorenz.leutgeb@radicle.xyz>
Do it by setting the environment variable 'CARGO_CMD' when running
'just', or by overriding the justfile variable 'cargo_cmd', like so:
CARGO_CMD='cargo --locked' just ...
or:
just cargo_cmd='cargo --locked' ...
Multiple commands take an RID as a command line argument. However, the
handling, which usually involves falling back to detection of the
repository associated with the current directory, is duplicated many
times over.
Introduce `fn rid_or_cwd` which can be used by many commands.
Many commands take RIDs as arguments. Most of them call the
corresponding struct member `repo`, some call it `rid`.
For consistency, name these members `repo` for all commands.
This change is intended to be invisible to the user. That is why,
in some cases, the "long" version of the argument must be renamed to
match `--rid`.
Daniel Norman
Josh Soref
Fintan Halpenny
Lorenz Leutgeb
:raw`
Lorenz Leutgeb
Adrian Duke
Richard Levitte