Inheco ODTC by cmoscy · Pull Request #894 · PyLabRobot/pylabrobot

cmoscy · 2026-02-15T19:34:24Z

ODTC backend. Runs on-device protocols, and uses a scratch protocol name for uploading and executing individual commands like set mount temperature.

…ution subclass

…er init Extract parsing utilities (_extract_dict_path, _extract_xml_parameter) and refactor 5 methods to use them. Add state verification after Initialize. Remove redundant logs and simplify error handling.

- Enforce method vs protocol conventions

…sistency Split interface into send_command (result) and start_command (handle); rename backend get_method_by_name/list_method_names to get_method/list_methods; standardize docstrings and update README/notebook.

point, backend implements ThermocyclerBackend) instead of the original custom layout; improves consistency with PyLabRobot and gives a clear recommended API. - Rename odtc_xml → odtc_model; remove odtc.py; add ODTCThermocycler as preferred resource. Use computed estimated duration (PreMethod 10 min, Method from steps); stop parsing duration from device. - Backend/thermocycler interface updates; pass estimated duration through all execution paths. - README: Recommended Workflows (run by name, round-trip for thermal performance, set_block_temperature); clarify preserve temps when reusing ODTC config (only durations/cycle counts safe to change); trim redundancy.

…tures/waits. Docs+Formatting

… viewing.

…olidation...

…ess. Implement Get_Step,Cycle,Remaining methods. Cleanup.

Stonelinks · 2026-03-16T00:23:02Z

Honestly, we are using it in a pretty simplistic fashion right now (single thread wait for a protocol to finish executing), so no haven't noticed any issues with client parallelism. Also very new to pylabrobot in general.

cmoscy · 2026-03-16T00:31:52Z

have you noticed the client side parallelism check being necessary? It adds some complexity to code. I am not opposed to including more client side checks when this is necessary for safe device operation, but this can also be a risk if we are more stringent than what the machines requires.

Don't think not including the client side check would break anything, totally fine if we want ODTC to just let the ODTC handle it by returning the error code 4 instead.

More as a source of documentation since the spec is disclosed here. But doesn't mean it has to be wired directly into execute_command

TD_SILA_FWCommandSet (25210).pdf

Move shared SiLA concepts (SiLAState, _current_state, _update_state_from_status, _complete_pending, _post_command, start_command, _on_http event dispatch) to InhecoSiLAInterface. ODTC overrides only what differs: _on_data_event, _complete_pending (lock/parallelism cleanup), send_command/start_command (state allowability, parallelism checks, lockId auto-injection). Delete _execute_command, PendingCommand, polling fallback, FirstEventType enum, COMMAND_FIRST_EVENT_TYPE dict, _validate_lock_id, _get_terminal_state. Net -550 lines. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

rickwierenga · 2026-03-16T00:33:27Z

prefer to have them handle it if it works, less for us to maintain :) error messages should be equally descriptive

cmoscy · 2026-03-16T00:34:52Z

prefer to have them handle it if it works, less for us to maintain :) error messages should be equally descriptive

Sure! Feel free to pull it out, or would you like me to make the corresponding changes/test?

rickwierenga · 2026-03-16T00:35:02Z

also, removed the ODTCThermocycler class for now, in anticipation of the refactor we are working on (http://discuss.pylabrobot.org/t/updating-plr-api-for-machine-interfaces-discussion/445)
moved your improvements of odtc sila into inheco sila for the shared stuff
only use 96 and 384 for types rather than duplicating

rickwierenga · 2026-03-16T00:36:33Z

Sure! Feel free to pull it out, or would you like me to make the corresponding changes/test?

no worries I can take care of it :) we have both the 96 and 384 well odtc. thanks for offering :)

…roperty Each backend method (reset, open_door, lock_device, etc.) now calls _run_async_command directly with ODTCCommand enum instead of routing through execute(). Remove ODTCCommand.EXECUTE_METHOD dispatcher; extract _execute_method_impl. Add event_receiver_uri property to base interface so reset() no longer needs the URI as a parameter. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…e odtc vars Remove variant field from ODTCHardwareConstraints (caller already knows it). Inline get_constraints. Default ODTCProtocol.name to SCRATCH_PROTOCOL_NAME so callers don't need fallback logic. Delete resolve_protocol_name function. Rename local `odtc` variables to `protocol` or `odtc_protocol` for clarity. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Delete odtc_method_to_protocol (ODTCConfig return unused), merge into odtc_protocol_to_protocol which now returns Protocol directly. Merge estimate_odtc_protocol_duration_seconds into estimate_method_duration_seconds. Inline _get_method_only_by_name, _stored_to_odtc_protocol, _read_opt_attr, list_method_names, list_premethod_names. Delete dead get_premethod_by_name. Rename odtc params to odtc_protocol in model.py. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

rickwierenga · 2026-03-16T05:21:18Z

what is the difference between ODTCMethodSet vs ProtocolList?

- _get_local_ip: reuse _get_link_local_interfaces from discovery for 169.254.x.x devices (UDP routing trick picks wrong interface on multi-homed hosts) - send_command: wrap await fut with asyncio.wait_for to prevent permanent hang when ResponseEvent never arrives - _setup_full_path: poll GetStatus until device leaves startup before sending Reset (device rejects Reset during boot) - Remove STARTUP from Reset allowed states (device rejects it) - _on_http: skip non-dict event payloads from soap_decode - Move SiLAError, SiLATimeoutError to base; ODTC imports from base - Remove duplicate SOAP_RESPONSE_ErrorEventResponse - _check_state_allowability raises directly with power-cycle hint for inError - Allow GetStatus/GetDeviceIdentification/Reset in error states Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

cmoscy · 2026-03-16T19:37:07Z

ProtocolList

Level of effort. ProtocolList is just going to pull stored method names for reference (or running an existing protocol by name). Where ODTCMethodSet will parse them into their respective ODTCProtocol objects (for detailed information, using as a starting point to edit and send as a new protocol, etc).

- _get_local_ip: reuse _get_link_local_interfaces for 169.254.x.x - HTTP server: always return 200 with SOAP response, never 500 (500 causes device to go inError) - _on_http: skip non-dict event payloads from soap_decode - send_command: 60s timeout for non-ExecuteMethod commands, lifetime_of_execution for ExecuteMethod only - Remove startup polling (Reset works in startup) - Allow Reset/GetStatus in all states - Move SiLAError/SiLATimeoutError to base class - _check_state_allowability raises with power-cycle hint for inError Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Remove _current_state, _update_state_from_status, and all local state transitions (IDLE↔BUSY, →INERROR, →ERRORHANDLING). The device reports its own state via StatusEvents and GetStatus — no need to mirror it. Add request_status() to InhecoSiLAInterface as the single method for querying device state (returns SiLAState). Move code-9 handling (state query + power-cycle hint) into the base class and delete the ODTC override. Update SCILABackend and ODTCBackend to delegate to it. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

PLR never sends PrepareForInput/PrepareForOutput — we always use OpenDoor/CloseDoor. Remove the normalization layer and the aliases from STATE_ALLOWABILITY. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…DTCMethodSet ProtocolList was always derived from ODTCMethodSet and list_protocols/list_methods were trivial wrappers around get_method_set. Moved get_method_by_name into ODTCMethodSet.get(), added __str__ for nice display. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…MMANDS send_command always waits now. Only execute_method uses asyncio.create_task for non-blocking execution. Simple commands (open_door, close_door, reset, etc.) no longer accept wait parameter. Timeout and lock tracking moved to callers instead of being hardcoded in send_command. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

_post_command now returns raw (decoded, return_code, message) without interpreting. send_command handles success codes (1, 2, 3) and delegates errors to _handle_error_code (renamed from _handle_return_code). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Removed _check_parallelism, _executing_commands tracking. Kept PARALLELISM_TABLE and STATE_ALLOWABILITY as reference documentation. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

send_command_async sends command and returns (future, request_id) without waiting. For sync commands (return_code 1), future is already resolved. send_command just awaits the future with a timeout. Lock/unlock and lockId injection moved to base InhecoSiLAInterface. ODTC send_command override and _complete_pending deleted. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

run_protocol, execute_method, run_stored_protocol all return None now. Backend tracks _current_request_id and _current_protocol directly. Callers use is_method_running(), get_current_step_index(), stop_method(), wait_for_method_completion() instead of a handle object. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…ify ODTCProtocol defaults - run_protocol calls set_block_temperature before executing method so device has matching start temperatures (required by ODTC hardware) - wait=True awaits the SiLA future instead of polling is_method_running (premethods hold temperature forever so polling never returns) - ODTCProtocol.datetime defaults to generate_odtc_timestamp(), name defaults to SCRATCH_PROTOCOL_NAME via protocol_to_odtc_protocol - XML parsers use elem.attrib[] for required fields (fail loud) - Removed **kwargs from open_lid, close_lid, set_block_temperature Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- Delete open_door/close_door (inline into open_lid/close_lid) - Delete wait_for_method_completion, wait_for_completion_by_time, _report_progress_once, _run_progress_loop_until (use TC.wait_for_profile_completion) - Delete _get_effective_lifetime, replace lifetime_of_execution with simple timeout param - Delete progress_log_interval/progress_callback from constructor - Delete SCRATCH_PROTOCOL_NAME constant, add is_scratch bool on ODTCProtocol - Simplify get_hold_time/get_*_index/get_*_count to return 0 when idle - Document backend spec: return 0 when no profile is running - Trim verbose docstrings on unsupported methods Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- Delete open_door/close_door, inline into open_lid/close_lid - Delete wait_for_method_completion and related progress machinery - Replace lifetime_of_execution with simple timeout parameter - Add is_scratch bool on ODTCProtocol, delete SCRATCH_PROTOCOL_NAME - execute_method takes ODTCProtocol directly - Delete _upload_odtc_protocol, merge into upload_protocol - upload_protocol only accepts ODTCProtocol - run_protocol accepts Protocol or ODTCProtocol, auto-runs premethod - Delete get_default_config, instantiate ODTCConfig directly - Freeze ODTCConfig, remove _validate flag and constraints property - Delete validate_volume_fluid_quantity (was validating its own output) - _current_protocol is always ODTCProtocol (no isinstance checks) - get_hold_time/get_*_index/get_*_count return 0 when idle - Backend spec: document return 0 when no profile running Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

rickwierenga · 2026-03-19T06:00:55Z

should we trust ODTCConfig or ODTCProtocol defaults?

rickwierenga · 2026-03-19T06:02:37Z

Level of effort. ProtocolList is just going to pull stored method names for reference (or running an existing protocol by name)

ODTCMethodSet could easily provide the list of names, so I deleted ProtocolList

rickwierenga · 2026-03-19T06:28:15Z

does build_progress_from_data_event typically get "current_step_index" etc. in the payload? what does this depend on? the conditional parsing smells suspicious

rickwierenga · 2026-03-19T06:30:41Z

in general do you know how rigid the data stream is? many dicts say .get rather than just direct access. I think we should be able to trust the data / if there are differences explain them and handle explicitly

…o base SiLA File split (model -> protocol -> xml, no circular imports): - odtc_model.py: dataclasses, constants, field metadata - odtc_protocol.py: protocol conversion, progress, stage/step helpers - odtc_xml.py: XML serialization and parsing Move DataEvent storage/retrieval to base InhecoSiLAInterface: - _data_events_by_request_id, _on_data_event, get_data_events - ODTC override adds progress logging only Other: - build_progress_from_data_event replaces ODTCProgress.from_data_event - _parse_data_event_payload raises on bad data instead of returning None - payload parameter is now required (not Optional) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

cmoscy added 30 commits February 10, 2026 11:08

ODTC Sila, Codec, Connection Init

278586b

Consolidate classes, async commands and rquest_id handling

1358d29

Add async command pattern: CommandExecution base class and MethodExec…

92143c6

…ution subclass

Doc and test notebook

813e569

States

d41c26a

refactor(odtc): centralize response parsing and verify idle state aft…

e8d3d3a

…er init Extract parsing utilities (_extract_dict_path, _extract_xml_parameter) and refactor 5 methods to use them. Add state verification after Initialize. Remove redundant logs and simplify error handling.

Formatting and notebook example

1f9417f

feat: Add protocol management and temperature control to Inheco ODTC

ad1a6f2

- Enforce method vs protocol conventions

Formatting and fluid quantity checking

94b4315

Sila2 Style Command Lifecycle Management

ade262d

ODTC: send_command vs start_command, get_method/list_methods, doc con…

cb86ea3

…sistency Split interface into send_command (result) and start_command (handle); rename backend get_method_by_name/list_method_names to get_method/list_methods; standardize docstrings and update README/notebook.

Merge branch 'PyLabRobot:main' into odtc

2bb0fea

Reconnect without Reset. Simulation Mode. Clean up Prints for tempera…

44ab52a

…tures/waits. Docs+Formatting

Merge branch 'PyLabRobot:main' into odtc

b4ad920

Remove redundant notebook

b94e165

Separate PreMethods vs Methods for Viewing. Display full protocol for…

ea6ceae

… viewing.

Checkpoint. Data Event handling and progress.

8371259

Refactor. Comms reorganization and ODTC Method, Protocol, Config Cons…

ca8d579

…olidation...

ODTCProtocol

939e436

Replaced StoredProtocol with ODTCProtocol

8d7ed52

Progress Polling

4c5a61b

Register execution ids for tracking

dc91515

Fix

ed389ec

wait is __await__

b35db2a

Remove prints

30a3d64

Centralize ODTCProtocol Lifecycle and Progress Reporting to ODTCProgr…

3dda9fe

…ess. Implement Get_Step,Cycle,Remaining methods. Cleanup.

Merge branch 'PyLabRobot:main' into odtc

fd07d35

Formatting

a824c92

Line Formatting

ab4264a

rickwierenga and others added 3 commits March 15, 2026 17:45

rickwierenga and others added 12 commits March 16, 2026 12:50

delete client-side parallelism enforcement, let device handle it

07bc94b

Removed _check_parallelism, _executing_commands tracking. Kept PARALLELISM_TABLE and STATE_ALLOWABILITY as reference documentation. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Inheco ODTC#894

Inheco ODTC#894
cmoscy wants to merge 63 commits intoPyLabRobot:mainfrom
cmoscy:odtc

cmoscy commented Feb 15, 2026

Uh oh!

Stonelinks commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026 •

edited

Loading

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026 •

edited

Loading

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

cmoscy commented Feb 15, 2026

Uh oh!

Stonelinks commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

rickwierenga commented Mar 16, 2026

Uh oh!

cmoscy commented Mar 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

rickwierenga commented Mar 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

cmoscy commented Mar 16, 2026 •

edited

Loading

cmoscy commented Mar 16, 2026 •

edited

Loading