Repository navigation
1.21.0
Groundwork for dive computers on more transports than Bluetooth LE: open a device on any iostream, download over a session with a commit-acknowledge fingerprint, identify computers the same way on every transport, and (macOS) custom, tapped and recorded streams. The iOS CBPeripheral download keeps its contract. Full details in CHANGELOG.md.
Added
DiveComputerSession(openon an iostream you opened, oradopt/adoptBLEan open device) andDiveLogRetriever.retrieveDiveLogs(session:...), which completes with aDiveDownloadResultand aFingerprintCommityou acknowledge once the dives are stored. Session downloads follow the same rejected-dive rule as BLE: a dive the parser rejects holds the fingerprint once and is skipped as unreadable the second time. BLE-adopted sessions share the manager's close deferral (#85)DiveComputerSessionLease: one open session at a time, BLE or cable (#85)DiveComputerIdentity,DiveComputerLocator,DCTransportandDeviceCatalog(every libdivecomputer descriptor at runtime, with canonical names); fingerprints and rejected dives can be keyed by identity, withmigrateLegacyKeys/DeviceStorage.migrateFingerprintKeys()to move older records over in place (#87)- macOS:
DCCustomStream,TapStream(cancellable within 100 ms),DCRecording(.dcrec) andScriptedStreamreplay;PacketEvent.channelnames the link a packet crossed (#86) - LibDCBridge:
open_device_on_iostream,find_descriptor_by_name_transport,libdc_set_cancel,libdc_close_device_data,libdc_cable_transports()andlibdc_has_custom_baud()(#82, #84, #85) - macOS: libdivecomputer builds with
IOSSIOSPEED, so non-standard serial rates (Sporasub SP2 460800, Cochran EMC 850000) work, plus serial latency and monotonic timeouts (#82) - DC Tester Mac: a sandboxed macOS probe for serial, USB-HID, raw USB, BLE and Garmin volume access, for the cable go/no-go matrix in
Docs/MAC_SANDBOX_PROBE.md(#80) - CI: macOS
swift test, generic iOS/macOS builds, iOS Simulator tests and the DC Tester build on every PR (#83)
Changed
DeviceConfiguration.DeviceFamilyhas a case for every libdivecomputer family (47, was 17). Existing cases and raw values are unchanged, but an exhaustiveswitchwithoutdefaultno longer compiles (#87)open_ble_deviceopens through the shared open helper, with the same 5 attempts 3 s apart (#84)- The
CBPeripheralretrieveDiveLogsoverload runs on the same download core as sessions; its behaviour is unchanged (#85)
Fixed
- Every downloaded dive is in
viewModel.diveswhencompletionruns; the last one used to land a main-queue turn later (#85) DiveDataViewModel.progressno longer slips back to.inProgressafter.completed,.noNewDives,.emptyRead,.cancelledor.failed, and a download's progress reset no longer clearslastHandshakeAcknowledged(#85)DeviceFingerprintStorageserializes its writes;DeviceStoragedrops an unreadable record on its own instead of emptying the list, andreconcileUUIDonly re-keys BLE records (#87)- Suunto Nautic/Ocean: a dive's start comes from
dc_datetime_mktime, so it stays the same instant once the parser reports each dive's UTC offset. No change at this release's libdivecomputer pin (#106, @aleskiontherun) find_descriptor_by_namereturnsDC_STATUS_INVALIDARGSfor a NULL argument instead of crashing (#84)
libdivecomputer is unchanged (still 1b201bb). The per-dive UTC offset (deepsealabs/libdivecomputer#19) comes in a later release, after a watch test.