Bump posthog-rs from 0.10.0 to 0.15.1

[dependabot]

Bumps posthog-rs from 0.10.0 to 0.15.1.

Release notes

Sourced from posthog-rs's releases.

0.15.1

Patch changes

• 91dfe67 Include canonical feature flag request fields for remote evaluation, including person/group properties, GeoIP disable, and flag key filtering. — Thanks @​marandaneto!

0.15.0

Minor changes

• 4294c51 Panic autocapture (opt-in). Set ErrorTrackingOptions::capture_panics to true and initialize the global client with init_global, and the SDK installs a process-wide std::panic hook that captures panics as personless $exception events through the global client, then calls the previously installed hook. Each event carries the panic payload, the panic-site location ($exception_panic_file/_line/_column), and a call-site stack trace honoring capture_stacktrace. Capture routes through the background worker thread, so it needs no async runtime and a panicking before_send hook can't abort the process. Gated by the default-on error-tracking feature.

  Panic autocapture is global-only: a panic hook is process-global (std::panic::set_hook), so it pairs with the process-global client. There is intentionally no per-Client panic API for now.

  New ErrorTrackingOptions: capture_panics (default false). The flush the hook performs on the panicking thread is bounded by a fixed, short timeout (2s) so a slow or unreachable PostHog can't freeze the crashing process or delay its panic message.

  Delivery is best-effort: the just-captured $exception is flushed ahead of queued retries, but under sustained backpressure (a slow/unreachable endpoint or a large capture backlog) it may not be sent within the bound before the process exits. — Thanks @​cat-ph!

0.14.3

Patch changes

• 447ed07 Respect Retry-After as minimum retry delay — Thanks @​marandaneto!
• 650746c Route feature flag called events through the normal capture transport. — Thanks @​marandaneto!

0.14.2

Patch changes

• 10810b1 Type-coerce capture-v1 options before placing them on the wire. A caller value whose type doesn't match the backend's strict Options schema is now coerced when possible, or dropped (backend applies its default) rather than rejecting the whole batch. No effect on default (v0) capture behavior. — Thanks @​eli-r-ph!

0.14.1

Patch changes

• e447b36 Harden the transport's in-flight event counter against underflow. The counter is decremented from several paths (before_send drops, partial v1 batch results, terminal outcomes, shutdown drops, channel drain); a decrement bug on any of them would previously underflow the AtomicUsize and wrap to a huge value, making the bounded queue look permanently full and silently dropping every subsequent event. Decrements now saturate at 0 (with a debug_assert to surface the bug in tests), so a release build degrades gracefully instead of wedging the queue. — Thanks @​eli-r-ph!

0.14.0

Minor changes

• 155d00a Runtime-independent background event transport. capture and capture_batch are now non-blocking enqueues onto a background worker — a plain std::thread with a blocking HTTP client, independent of any async runtime — that batches events, retries transient failures with backoff (honoring Retry-After), and sends them. They no longer block on the network or return delivery errors.

  New public API: flush() (awaited on the async client, blocking on the blocking client), shutdown() (flush + stop the worker + join; idempotent; drops further captures), and flush-on-Drop. New ClientOptions: flush_at, max_batch_size, flush_interval_ms, max_queue_size (a bounded queue that drops with a single warning when full), and shutdown_timeout_ms (bounds the shutdown/Drop drain). before_send hooks now run on the worker thread, so they apply to every queued event.

  Breaking change (0.x): capture and capture_batch no longer return a Result — and are no longer async on the async client. They enqueue the event and return immediately (infallibly) instead of awaiting delivery; transient HTTP failures surface as logged warnings rather than Err. Call flush() or shutdown() before process exit to ensure queued events are delivered. — Thanks @​cat-ph!

0.13.3

Patch changes

• 911ceae Internal changes to the capture-v1 feature path. No effect on default (v0) capture behavior. — Thanks @​eli-r-ph!

0.13.2

Patch changes

• a4d9724 Remove duplicated per-event API keys from v0 batch payloads. — Thanks @​marandaneto!

... (truncated)

Changelog

Sourced from posthog-rs's changelog.

0.15.1 — 2026-06-30

Patch changes

• 91dfe67 Include canonical feature flag request fields for remote evaluation, including person/group properties, GeoIP disable, and flag key filtering. — Thanks @​marandaneto!

0.15.0 — 2026-06-29

Minor changes

• 4294c51 Panic autocapture (opt-in). Set ErrorTrackingOptions::capture_panics to true and initialize the global client with init_global, and the SDK installs a process-wide std::panic hook that captures panics as personless $exception events through the global client, then calls the previously installed hook. Each event carries the panic payload, the panic-site location ($exception_panic_file/_line/_column), and a call-site stack trace honoring capture_stacktrace. Capture routes through the background worker thread, so it needs no async runtime and a panicking before_send hook can't abort the process. Gated by the default-on error-tracking feature.

  Panic autocapture is global-only: a panic hook is process-global (std::panic::set_hook), so it pairs with the process-global client. There is intentionally no per-Client panic API for now.

  New ErrorTrackingOptions: capture_panics (default false). The flush the hook performs on the panicking thread is bounded by a fixed, short timeout (2s) so a slow or unreachable PostHog can't freeze the crashing process or delay its panic message.

  Delivery is best-effort: the just-captured $exception is flushed ahead of queued retries, but under sustained backpressure (a slow/unreachable endpoint or a large capture backlog) it may not be sent within the bound before the process exits. — Thanks @​cat-ph!

0.14.3 — 2026-06-29

Patch changes

• 447ed07 Respect Retry-After as minimum retry delay — Thanks @​marandaneto!
• 650746c Route feature flag called events through the normal capture transport. — Thanks @​marandaneto!

0.14.2 — 2026-06-27

Patch changes

• 10810b1 Type-coerce capture-v1 options before placing them on the wire. A caller value whose type doesn't match the backend's strict Options schema is now coerced when possible, or dropped (backend applies its default) rather than rejecting the whole batch. No effect on default (v0) capture behavior. — Thanks @​eli-r-ph!

0.14.1 — 2026-06-24

Patch changes

• e447b36 Harden the transport's in-flight event counter against underflow. The counter is decremented from several paths (before_send drops, partial v1 batch results, terminal outcomes, shutdown drops, channel drain); a decrement bug on any of them would previously underflow the AtomicUsize and wrap to a huge value, making the bounded queue look permanently full and silently dropping every subsequent event. Decrements now saturate at 0 (with a debug_assert to surface the bug in tests), so a release build degrades gracefully instead of wedging the queue. — Thanks @​eli-r-ph!

0.14.0 — 2026-06-24

Minor changes

• 155d00a Runtime-independent background event transport. capture and capture_batch are now non-blocking enqueues onto a background worker — a plain std::thread with a blocking HTTP client, independent of any async runtime — that batches events, retries transient failures with backoff (honoring Retry-After), and sends them. They no longer block on the network or return delivery errors.

  New public API: flush() (awaited on the async client, blocking on the blocking client), shutdown() (flush + stop the worker + join; idempotent; drops further captures), and flush-on-Drop. New ClientOptions: flush_at, max_batch_size, flush_interval_ms, max_queue_size (a bounded queue that drops with a single warning when full), and shutdown_timeout_ms (bounds the shutdown/Drop drain). before_send hooks now run on the worker thread, so they apply to every queued event.

  Breaking change (0.x): capture and capture_batch no longer return a Result — and are no longer async on the async client. They enqueue the event and return immediately (infallibly) instead of awaiting delivery; transient HTTP failures surface as logged warnings rather than Err. Call flush() or shutdown() before process exit to ensure queued events are delivered. — Thanks @​cat-ph!

0.13.3 — 2026-06-23

Patch changes

... (truncated)

Commits

• d1bbe85 chore: Release v0.15.1 [skip ci]
• 91dfe67 chore: update SDK compliance harness to 0.8.0 (#163)
• a9e518f chore: Release v0.15.0 [skip ci]
• 4294c51 feat: panic autocapture (#157)
• a178b8c chore: Release v0.14.3 [skip ci]
• d082ddd fix(flags): retry flag requests on transient network errors (#156)
• 650746c fix: route flag called events through capture transport (#162)
• 447ed07 fix: respect Retry-After minimum delay (#161)
• cf34819 chore: exclude test harness APIs from public API snapshot (#164)
• 8263e0b chore: Release v0.14.2 [skip ci]
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)