With #12 we also found the following "missing" features; none are super important, but could be nice-to-haves as part of the demo.
- External Crash reporter — with
make crash-reporter we bundle this in, but currently the user feedback seems to not be working (returns "Response status code does not indicate success: 403 (Forbidden)."). This might also be an issue with the external reporter itself.
- Offline caching / cache keep (
sentry_options_set_cache_keep, set_cache_max_items/size/age) — Add a "Go Offline" toggle in the Settings panel that simulates network loss. Events would queue locally and drain when toggled back. Very visual for demos. Implementation: the SDK handles this internally once the options are set; we'd just need to enable SENTRY_CACHE_KEEP_ALWAYS and expose a UI toggle.
- HTTP retry (
sentry_options_set_http_retry) — Pairs naturally with offline caching above. Enable it in sentry_manager.cpp alongside the cache options. One line in init(). Could show retry behavior in the console log when the backend is unreachable.
- before_screenshot hook (
sentry_options_set_before_screenshot) — Could wire this up in sentry_manager.cpp to stamp the screenshot with an overlay (e.g. timestamp or scenario name) before capture. Only fires on Windows where SDK screenshots are enabled. Niche value.
- before_send_log / before_send_metric (
sentry_options_set_before_send_log, sentry_options_set_before_send_metric) — Filter hooks for logs and metrics. Could add a Settings toggle to suppress certain log levels or metric names, showing how an app can gate telemetry at runtime. Implementation in sentry_manager.cpp alongside before_send.
- Attachment types (
sentry_attachment_set_type, SENTRY_ATTACHMENT_TYPE_*) — The calibration blob in the convoluted chain scenario currently uses plain sentry_attach_bytes. Could call sentry_attachment_set_type to mark it as a specific type for richer display in the Sentry UI. Small change in chaos.cpp.
- Attachment content type (
sentry_attachment_set_content_type) — Set MIME types on attachments (calibration blob as application/octet-stream, screenshot as image/png). Pairs with attachment types above.
- User consent management (
sentry_user_consent_give/revoke/get) — Add a GDPR-style consent toggle in Settings. When revoked, events are cached but not sent; when given, the backlog drains. Pairs with offline caching (nr.1). Would need a consent state in the UI and calls in sentry_manager.cpp.
- Large attachment uploads (
sentry_options_set_enable_large_attachments) — Experimental TUS-based upload for large files. The app-hang scenario already attaches a 256 KB blob; could bump that to several MB and enable this option to show it working. One-line option in sentry_manager.cpp, size bump in chaos.cpp.
on_crash callback (sentry_options_set_on_crash) — Separate hook that fires only on crashes, replacing before_send for crash events. Could use this for crash-specific enrichment while keeping before_send for non-crash events. Good teaching example for the before_send / on_crash split.- Crashed-last-run detection (
sentry_get_crashed_last_run) — Check at startup whether the previous run crashed. Could show a "last session crashed" banner in the UI. Very visual and easy to implement — one call right after sentry_init.
dist option (sentry_options_set_dist) — Set to "macos" / "linux" / "windows" so Sentry can distinguish OS builds of the same release. One-liner in sentry_manager.cpp using the existing EMPOWER_PLATFORM macro.- Client reports (
sentry_options_set_send_client_reports) — Track discarded/filtered events in Sentry. One-liner in sentry_manager.cpp. Pairs well with the filtering callbacks above — shows what got dropped.
sentry_metrics_count — The demo uses gauge and distribution but not count. Adding a counter (e.g. "scenarios triggered") would complete the metrics trifecta. Small change in chaos.cpp.- Programmatic user feedback (
sentry_capture_feedback_with_hint) — A "Send Feedback" button in the UI, capturing feedback via the API rather than only through the external crash reporter. Would need a small text-input dialog in the GUI.
- Local scopes (
sentry_local_scope_new + sentry_capture_event_with_scope) — Per-event scope enrichment without polluting the global scope. The chaos scenarios currently set tags globally; local scopes would be cleaner and demonstrate the API.
- Transaction filtering (
sentry_options_set_before_transaction) — Filter or modify transactions before send. Could add a Settings toggle to drop certain transaction types, showing how an app can control which traces are sent.
With #12 we also found the following "missing" features; none are super important, but could be nice-to-haves as part of the demo.
make crash-reporterwe bundle this in, but currently the user feedback seems to not be working (returns "Response status code does not indicate success: 403 (Forbidden)."). This might also be an issue with the external reporter itself.sentry_options_set_cache_keep,set_cache_max_items/size/age) — Add a "Go Offline" toggle in the Settings panel that simulates network loss. Events would queue locally and drain when toggled back. Very visual for demos. Implementation: the SDK handles this internally once the options are set; we'd just need to enableSENTRY_CACHE_KEEP_ALWAYSand expose a UI toggle.sentry_options_set_http_retry) — Pairs naturally with offline caching above. Enable it insentry_manager.cppalongside the cache options. One line ininit(). Could show retry behavior in the console log when the backend is unreachable.sentry_options_set_before_screenshot) — Could wire this up insentry_manager.cppto stamp the screenshot with an overlay (e.g. timestamp or scenario name) before capture. Only fires on Windows where SDK screenshots are enabled. Niche value.sentry_options_set_before_send_log,sentry_options_set_before_send_metric) — Filter hooks for logs and metrics. Could add a Settings toggle to suppress certain log levels or metric names, showing how an app can gate telemetry at runtime. Implementation insentry_manager.cppalongsidebefore_send.sentry_attachment_set_type,SENTRY_ATTACHMENT_TYPE_*) — The calibration blob in the convoluted chain scenario currently uses plainsentry_attach_bytes. Could callsentry_attachment_set_typeto mark it as a specific type for richer display in the Sentry UI. Small change inchaos.cpp.sentry_attachment_set_content_type) — Set MIME types on attachments (calibration blob asapplication/octet-stream, screenshot asimage/png). Pairs with attachment types above.sentry_user_consent_give/revoke/get) — Add a GDPR-style consent toggle in Settings. When revoked, events are cached but not sent; when given, the backlog drains. Pairs with offline caching (nr.1). Would need a consent state in the UI and calls insentry_manager.cpp.sentry_options_set_enable_large_attachments) — Experimental TUS-based upload for large files. The app-hang scenario already attaches a 256 KB blob; could bump that to several MB and enable this option to show it working. One-line option insentry_manager.cpp, size bump inchaos.cpp.on_crashcallback (sentry_options_set_on_crash) — Separate hook that fires only on crashes, replacingbefore_sendfor crash events. Could use this for crash-specific enrichment while keepingbefore_sendfor non-crash events. Good teaching example for thebefore_send/on_crashsplit.sentry_get_crashed_last_run) — Check at startup whether the previous run crashed. Could show a "last session crashed" banner in the UI. Very visual and easy to implement — one call right aftersentry_init.distoption (sentry_options_set_dist) — Set to"macos"/"linux"/"windows"so Sentry can distinguish OS builds of the same release. One-liner insentry_manager.cppusing the existingEMPOWER_PLATFORMmacro.sentry_options_set_send_client_reports) — Track discarded/filtered events in Sentry. One-liner insentry_manager.cpp. Pairs well with the filtering callbacks above — shows what got dropped.sentry_metrics_count— The demo uses gauge and distribution but not count. Adding a counter (e.g. "scenarios triggered") would complete the metrics trifecta. Small change inchaos.cpp.sentry_capture_feedback_with_hint) — A "Send Feedback" button in the UI, capturing feedback via the API rather than only through the external crash reporter. Would need a small text-input dialog in the GUI.sentry_local_scope_new+sentry_capture_event_with_scope) — Per-event scope enrichment without polluting the global scope. The chaos scenarios currently set tags globally; local scopes would be cleaner and demonstrate the API.sentry_options_set_before_transaction) — Filter or modify transactions before send. Could add a Settings toggle to drop certain transaction types, showing how an app can control which traces are sent.