--hmac "" silently produced an empty-byte HMAC key at the protocol
boundary, so an accidentally-empty shell expansion (--hmac "$VAR")
could look like auth was deliberately enabled while actually using a
predictable empty key. Reject a zero-length --hmac value at parse
time in the client CLI, the TUI client (via the shared
CommonClientArgs), and the server CLI. Non-empty values, including
whitespace-only ones, are accepted exactly as supplied.
`SampleMode::Exact` retained exact samples for eight of the eleven public
timing metrics. Send call, timer error, and server processing always built
their `TimeMetric` with retention disabled, so a finite run reported no
median for them even though they are the same public `TimeStats` type and
appear in the same summary table with a Median column.
Derive one `retain_exact` decision from the sample mode in `CoreStats::new`
and apply it to every timing metric, so exact mode means the same thing for
all of them. Running-only mode is unchanged and still retains none of them.
The retention estimator counted eight exact sample streams per measurable
reply. Rename it to the eleven exact samples an ordinary successful probe
can retain and document the wider optional-measurement upper bound, which
raises the finite estimate on this target from 668 to 764 bytes per probe.
A test counts the metrics a fully measured probe actually retains and
compares it against that constant, so retention and its estimate cannot
drift apart unnoticed.
No new median-capability type was needed: `TimeStats::median_ns` already
distinguishes a retained median from an absent one, and an optional
measurement the server never supplied still produces no samples at all.
Both client applets duplicated the step from parsed arguments to a runnable
managed setup: target specs, managed target preparation, a ClientConfig
built from an arbitrary primary target, pacing conversion, and an identical
ManagedClientConfig skeleton with its own copy of the event-capacity
constant. The primary-target step was the worst of it, since the managed
driver replaces server_addr per target, so the address the applets went out
of their way to find was immediately discarded.
shared::client::prepare now performs that step once: it validates the target
set first, then returns the validated targets, one target-free ClientConfig
template, and the pacing, with managed_config()/managed_targets() for the
driver. CommonClientArgs::to_client_config no longer takes a server address.
Applet-specific semantics are unchanged: each command keeps its own duration
default and help text, the client still accepts no target for
--list-columns, and the TUI still rejects an empty target list at parse
time.
Stream output, the run summary, and the TUI each carried their own copy of
the same scalar vocabulary, and they had drifted: the TUI spelled
microseconds "us" while the others used "µs", percentages guarded
non-finite values in one place but not another, and the absent-value
placeholder was spelled inline everywhere.
cmd::format now owns nanosecond-scalar, percentage, and count formatting
with one documented unit-selection and rounding policy, µs everywhere, and
signed values preserved. The frontends keep their own metric sets and
layouts; the TUI keeps its coarse wall-clock span policy, now named
format_span and documented as deliberately distinct, and uses the shared
scalar vocabulary below one second.
classify_client_error now matches every ClientError variant explicitly, so a
new variant fails to compile here until its real classification is chosen.
NoTokioRuntime classifies as InvalidState and DatagramLengthMismatch as
Socket; ManagedTargetFailureKind::Other survives only as a deliberate
category with no default path into it.
ManagedTargetFailurePhase and ManagedTargetFailureKind gain controlled
Display impls, and the CLI client/TUI failure diagnostics use them instead
of leaking Debug enum spelling into user-facing text.
Remove the optional "stats" feature by merging it into the "client" and "tui" features to reduce conditional compilation complexity. Replace the "client-runtime" internal feature with direct dependency sets.
Add a scheduler-owned saturating peer-close counter shared with the group session. Update it before best-effort TargetFinished publication so bounded subscriber overflow cannot hide the live shutdown condition, while retaining the existing events, bounded history, and aggregate accounting.
Make grouped CLI and TUI supervision consult the durable count on each existing activity loop and funnel both count and event observations through one-shot stop policy. Preserve interruption suppression, queued-event draining, structured failures, drop reporting, and peer-close diagnostic precedence.
Cover scheduler-only accounting, non-peer exclusions, receiver and removal races, deterministic capacity-one DropOldest eviction with an active target, frontend one-shot policy, and the existing process-level peer-close and interruption behavior.
Replace the manual single-target ClientSession loops in the stream and TUI adapters with ManagedClient::start_with_subscription so the existing worker owns opening, deadline-aware pacing, receive handling, timeout polling, draining, and close.
Use bounded subscription waits for interruption and input polling, drain queued events before and after join, and report per-subscription drops without changing output, statistics, labels, summaries, warning, or peer-close policy. Add process and unit regressions for 10 ms cadence, interruption, final-event draining, TUI terminal state, and overflow visibility.
Record a continuous peer-closed TargetFinished event as the reason for requesting group shutdown and stop each CLI/TUI group at most once. Drain the managed subscription and join before applying exit policy so queued output and target failures are preserved.
Give peer closure precedence over the cancellation caused by that internal stop while keeping explicit interruption and finite-run behavior distinct. Add bounded process regressions for partial closure, mixed failures, queued rows, and interruption plus focused TUI policy coverage.
Keep PeerClosed as a distinct successful library outcome while treating it as an error at the CLI and TUI boundary for uninterrupted continuous runs.
Track aggregate peer-close outcomes independently of bounded recent history and cover finite, continuous, grouped, mixed-failure, interruption, and TUI status behavior.
Track a monotonic discarded-event count per bounded subscription for all overflow policies, including after hub sealing or disconnection. Preserve nonblocking publication and existing queue policies.
Warn once in streaming CLI output and mark TUI summaries/status as incomplete when DropOldest discarded events. Add exact overflow-counter and incomplete-marker regressions.
Retain only the 256 most recent target outcomes in the final group snapshot while tracking total, successful, failed, and discarded outcome counts. Publish each authoritative TargetFinished event before inserting into the bounded history.
Update CLI/TUI exit decisions to use aggregate counts and add a dynamic churn regression proving every outcome remains observable while join retention stays bounded.
Track desired membership separately from active target state and make natural completion an explicit group policy. Empty authoritative updates now remove all targets without terminating the scheduler, while static CLI groups retain all-target completion.
Document that construction still requires an address-family-defining target and add regressions for empty replacement, later re-addition, idle control responsiveness, idle cancellation, and finite static completion.
Consume structured per-target outcomes in the streaming CLI and TUI so mixed failures remain visible while successful targets still produce output. Return a failure status when every target fails, and keep interruption behavior distinct.
Add strict-timeout integration coverage for mixed continuous completion and all-open-failure exit status, plus focused TUI outcome-state tests.
Extend the managed group stream with authoritative per-target terminal outcomes and structured failure classifications. Final client events are published first, followed by exactly one TargetFinished outcome for success, peer close, no-test, opening/runtime failure, removal, or cancellation.
Scheduler-owned cleanup now stops the receiver and seals subscriptions on natural completion, eliminating the event-consumer/join circular wait while preserving queued drain. TargetEventSubscription now yields ManagedGroupEvent, so consumers must handle Client and TargetFinished variants.
Strict-timeout regressions cover mixed success/failure, all-open failure, runtime failure, no-test ordering, removal, cancellation, and disconnection before join.
Enable simultaneous probing of multiple targets within the TUI using a new target status table and multi-series RTT graph. Refactor target resolution logic into a shared module to support consistent behavior across CLI and TUI components.
Add unit tests for target normalization and address family filtering, and expand integration tests to verify column rendering and CSV output for multiple targets. Update subscriber config to drop oldest events under backpressure for better CLI stability.
Enable concurrent probing of multiple targets using positional arguments or the `--target` flag with unique labels. Update output formats to include a `target` column for logical attribution in multi-target runs.
Simplify the default table view by moving verbose columns to the verbose mode and allow users to explicitly request the default layout using `--columns default`.
Replace fixed output modes with customizable table, CSV, TSV, and JSON Lines formats. This allows users to select specific data columns and control header visibility for better integration with scripts and monitoring tools.
Update the machine-readable wall clock field to "event_wall_ns" for consistency and restrict visibility of internal rendering types. The extensive test suite was removed as part of this cleanup.
Nils Andreas Svee