Repository navigation
fix(ohme): leave the Intelligent car slots to Octopus when the Intelligent device is the car - #5405
Merged
Merged
Conversation
…igent device is the car With ohme_automatic on and ohme_automatic_octopus_intelligent unset, the Ohme component took the Intelligent car slots for itself on any Intelligent tariff. Where the device Octopus Intelligent drives is the car and not the Ohme, Octopus schedules the charge through the car, and Ohme's session has nothing to do with the dispatches (#5402). Auto-detection now looks at the Octopus component's live, non-suspended Intelligent devices. The Ohme takes the slots when it is one of them, or when the account has no Intelligent device at all. When Octopus drives another device the slots are left to the Octopus component, Ohme's own schedule is not used as the plan, and ohme_control stands down. With every device suspended there are no dispatches, so Ohme's schedule is the plan at the tariff rate. The Octopus component now re-wires the car slots when their owner changes, so they are picked up when Ohme gives up a claim it made before the device was known. Fixes #5402 Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
Ownership release can incorrectly erase Ohme’s schedule for suspended devices or restore stale dispatch wiring after leaving an Intelligent tariff.
Review effort: Balanced
Findings: 1
What changed in this PR
Updates Ohme/Octopus Intelligent coordination so charging slots follow the device controlled by Octopus.
Changes:
- Detects whether Octopus controls the Ohme charger or another device.
- Rewires slot ownership when the controlled device changes.
- Adds regression tests and documentation.
| File | Description |
|---|---|
apps/predbat/ohme.py |
Selects slots based on the Intelligent device. |
apps/predbat/octopus.py |
Reconfigures slots after ownership changes. |
apps/predbat/tests/test_ohme.py |
Tests device detection and transitions. |
apps/predbat/tests/test_octopus_intelligent_devices.py |
Tests ownership-based rewiring. |
docs/car-charging.md |
Documents slot-selection behavior. |
docs/components.md |
Clarifies Ohme auto-detection. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
… to wire them to Review feedback on #5405. The Ohme component also gives up its claim on the car slots when there are no dispatches to be had from anyone - the Intelligent device suspended, or the tariff no longer Intelligent - and then wires them to the charger's own schedule. The owner-change re-wire added in this PR ran automatic_config() in those cases too: - with every device suspended it cleared the wiring Ohme had just made; - off an Intelligent tariff, where the device list is not refreshed and still holds the old device, it overwrote it with stale dispatch entities. The re-wire now only runs when the slots were released and there is a live, non-suspended device on a tariff that is still Intelligent; otherwise the new owner is just noted. automatic_config() also only clears car slot wiring that is still what it last put there, as the device-set change that follows a suspension would otherwise clear the same wiring one refresh later. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
chalfontchubby
added a commit
that referenced
this pull request
Oct 6, 2026
…alone when Octopus drives the car, unless asked to follow it When the car itself is the Octopus Intelligent device, the car starts and stops its own charge. Predbat now leaves the charger alone by default, handing over a charger it held without starting a charge, as it does when Octopus drives the charger. The new expert-mode switch, default off, has Predbat drive the charger to the car's dispatches instead, so the car cannot also charge on its own timers. The Ohme follows the same switch: with it on, ohme_control can run while Octopus drives the car, where #5405 otherwise stands it down. A change of the switch is picked up on the next Ohme poll. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
springfall2008
pushed a commit
that referenced
this pull request
Oct 12, 2026
…e and guest charging for GE, Ohme and Zappi (folds #5284 + #5311) (#5448) * feat(car): add a shared Predbat-led EV charger control mixin Ohme, myenergi and GivEnergy EVC each carry their own copy of the same control loop. CarChargerControl owns it once: read the per-car plan, charge inside a window and stop outside one, re-apply on drift, and release on read only or when the control switch is turned off, with the switch persisted across restarts. Components supply only the charger-specific send, release, connected and drift checks. Nothing uses it yet; the components move onto it in following commits. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * refactor(gecloud): drive EV chargers through the shared charger control GivEnergy EVC control moves onto CarChargerControl. The component keeps what is GivEnergy's own: which chargers exist and in what order, whether a car is connected, and the start-charge / stop-charge commands. Behaviour is unchanged: the switch keeps its storage key, chargers without a car are still left alone, and release still only restarts a charger Predbat had stopped. The EVC read only check now falls back to the set_read_only arg, like the other chargers, rather than the switch entity; the fallback only applies before Predbat sets the attribute at startup. The test double's get_arg answers set_read_only from the same flag as its switch, as Predbat does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * refactor(myenergi): drive Zappis through the shared charger control Zappi control moves onto CarChargerControl. The component keeps what is myenergi's own: Fast inside a window and Stopped outside one, drift detected from the polled mode, and release back to the saved mode or Eco+. One deliberate change: a Zappi with no car to follow (more Zappis than num_cars, briefly true while auto-config's num_cars reaches the base) is now left alone rather than Stopped, matching what GivEnergy EVC control already did. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * refactor(ohme): drive the charger through the shared charger control Ohme control moves onto CarChargerControl. The component keeps what is Ohme's own: max charge inside a window and pause outside one, drift detected from the session mode, and release back to Ohme's smart schedule with the user's target percent restored. Ohme has no control switch, so read only stays the only release. Its inline plan parser goes too, in favour of the shared one. That fixes a New Year case the copy missed: read just after midnight on 1 January, a window that started on 31 December parsed as the following December, so a charge running across midnight was paused. A release Ohme refuses is now retried on the next control cycle rather than attempted once, so a paused car is not stranded by one failed call. The control tick is wrapped the way myenergi's already is: a refused command logs a warning instead of failing the whole run, which would otherwise repeat every minute while read only stays on. The mixin now checks for a switch before touching storage, so a component with no switch never needs a storage attribute. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): address #5284 review - release chargers whose car has gone, retry refused GE commands - A charger Predbat holds that falls beyond the cars it follows (num_cars dropped) was never commanded again, left charging or stopped with nobody in control. apply() now releases it. - GivEnergy's async_send_evc_command() reports a refusal by returning None, so a refused release was taken as done and a stopped charger stranded - the failure the Ohme and myenergi retries already close. GE charger control now raises EVCCommandFailed on a refusal, so the command is not recorded and is tried again next cycle; GE's run loop catches it as a warning, as Ohme's and myenergi's already do. A refused start or stop is retried the same way rather than recorded as sent. - Docstring for a nested test helper. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): one charger refusing no longer blocks the others, and a released Zappi forgets its mode Found by a later review, in code from this PR: - Releasing, releasing a charger whose car has gone, and driving from the plan ran as one loop, so the first charger to refuse a command stopped every charger after it - and each cycle retried in the same order, so they stayed blocked. With two Zappis held Stopped and read only on, a refusing Zappi 1 left Zappi 2 Stopped for good. Each charger is now handled on its own; the first failure is still raised after the rest have had their turn, for the run loop to log and retry. A charger is forgotten as soon as its own release goes through, so a retry re-sends only to the one that refused. - A Zappi released on its own (its car gone when num_cars dropped) kept its saved mode, so taking it back later saved nothing new and the next release restored the stale mode. The saved mode is now dropped once that Zappi's release goes through. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(car): leave a charger to Octopus only when Octopus drives the charger Charger control now decides per car whether Octopus Intelligent is in charge of the charger. Where Octopus delivers its dispatches by driving the charger itself, Predbat leaves it alone - releasing it if it was holding it - so the two never fight. Where Octopus drives the car, or there is no Intelligent car, Predbat drives the charger from the car slot sensor, which then carries the Octopus dispatches, so the car cannot charge outside them on its own timers. Octopus car N is the Nth wired Intelligent device, and the Octopus component already records whether each is a charge point. When that cannot be told - slots from outside the Octopus component, or before it has discovered its devices - the charger is left alone, unless the component's control setting is explicitly true. Nothing turns control on by default yet; that follows. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(gecloud): turn EV charger control on with ge_cloud_automatic_evc ge_cloud_evc_control no longer defaults to false. Left unset, control follows ge_cloud_automatic_evc: setting that is the user asking Predbat to plan for the car, and it is what maps each charger to its car. An explicit false keeps control off, and only an explicit true warns when a prerequisite is missing. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(ohme): turn charger control on with ohme_automatic ohme_control no longer defaults to false. Left unset, control follows ohme_automatic, which registers the car. Where the Intelligent slots come from the Ohme, Octopus drives the charger and Predbat still leaves it alone - logged as Info for an unset control, and as a warning only when ohme_control was explicitly set true. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(myenergi): turn Zappi control on when the user wrote myenergi_automatic myenergi_zappi_control no longer defaults to false. Left unset, control turns on only when the user wrote myenergi_automatic or myenergi_automatic_zappi into apps.yaml themselves. Both default on, so following their values alone - the rule GE and Ohme use - would switch every Zappi user to Predbat control on upgrade, and a Zappi held off between windows can no longer divert solar in Eco/Eco+. The options considered are recorded in enable_control()'s docstring for review. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(car): charger control turns on with automatic setup, and Octopus Intelligent A new "Predbat starting and stopping your charger" section in car-charging.md sets out when each charger's control turns on while its setting is unset, the upgrade change for existing automatic users, and what happens on Octopus Intelligent depending on whether Octopus drives the charger or the car. The per-setting entries in apps-yaml.md and components.md now describe the unset default instead of false. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): judge Octopus per car from its own dispatch sensor, and never start a charge handing over Found by review before opening the stage 2 PR: - Read is_charger from the car's own wired octopus_intelligent_slot sensor instead of pairing car N with the Nth sorted Octopus device. Both the Octopus and Kraken components publish it there, so E.ON/EDF SmartFlex users are covered and a slot list in a different order can no longer pair a car with another car's device. - Only wait for Octopus discovery when the Octopus component runs its automatic setup. With octopus_automatic off it never wires the slots, so control never turned on. "Cannot tell yet" also gets its own log line rather than claiming the car is on IOG. - Handing a charger to Octopus no longer starts it. A normal release undoes a stop so a car is never stranded, but here Octopus decides when to charge: a GE charger Predbat stopped is left for Octopus, an Ohme has max charge turned off before it is resumed into its smart schedule, and a Zappi returns to its own mode, Eco+ in place of Fast. - The control settings have no default, which also skipped get_arg()'s boolean conversion, so a quoted "false" or a 0 read as set. parse_control_setting() normalises them to None, True or False in each component. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(ohme): only turn control on by default once the car's size and limit are set Predbat drives an Ohme with max charge, which overrides the target set in the Ohme app, so car_charging_battery_size and car_charging_limit are all that stop a charge. Left at their defaults (100 kWh, 100%) the car would be charged to full. Unset ohme_control now waits for both to be in apps.yaml and logs why; an explicit ohme_control: true is still the user's call. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(car): Ohme size and limit guard, Kraken, and handing over to Octopus Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): a charger Octopus drives stays with Octopus, and a car Octopus drives follows its dispatch sensor - switch.predbat_octopus_intelligent_charging only changes Predbat's own planning: Octopus goes on switching a charger that is its Intelligent device whatever the switch says. The device is now judged first, so turning the switch off no longer has Predbat fight Octopus for that charger. For a car Octopus drives, the switch still decides whether the charger follows the dispatches or Predbat's own plan. - Where Octopus drives the car, the charger also runs while the car's dispatch sensor is on. The plan carries the dispatches but is only republished every 5 minutes and leaves out a dispatch Octopus has not given any energy yet, so the charger held the car off for the first minutes of a new dispatch. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(car): guest charging switch for chargers Predbat drives While Predbat drives a charger it holds it off outside the planned windows, so a visitor's car could not charge - and stage 2 turns control on by default. Each component (GE, myenergi, Ohme) now publishes switch.predbat_<component>_guest_charging beside its control. On, Predbat releases the charger as it does for read only mode, so it runs on its own settings; how the guest charges is chosen on the charger itself. It turns itself off when a connected car is unplugged, on chargers that can tell (GE), and otherwise after GUEST_CHARGING_MAX_HOURS (12). It is deliberately not persisted, so a restart puts Predbat back in charge. The guest's charging is still counted as the car's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(car): end guest charging when the car is unplugged from a Zappi or Ohme Both report whether a car is plugged in - the Zappi's plug status, the Ohme's UNPLUGGED status - so guest charging now ends on unplug for every charger, not just GE. A new charger_control_car_plugged() hook carries it, separate from charger_control_connected(), so how Predbat drives an empty Zappi or Ohme is unchanged. An unknown plug state reads as plugged in, leaving the 12 hour limit. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): guest charging survives the owner unplugging, and dispatch ends are judged by the clock Found by review before opening the stage 2 PR: - Guest charging ended on any unplug, including the owner's car being unplugged to make way for the guest. Only a car plugged in after the switch went on now ends it (empty, plugged in, empty again); a guest who plugged in first is left to the 12 hour limit or a manual off. - A GE status blip (offline, unavailable, unknown) read as an unplug and ended guest charging. Only "available" and "idle" count as empty now, as the Zappi and Ohme already treat unknown states as plugged in. - The charger followed the dispatch sensor's on/off state, which only refreshes every couple of minutes and so ran the charger past the end of a dispatch. The dispatch times on the sensor are now judged against the clock, in both the Octopus and the Kraken formats, falling back to the state for a sensor with no times. - A toggle on the guest switch now flips it rather than turning it off. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(car): warn when Predbat cannot tell whether Octopus drives a charger A dispatch sensor that does not say which device Octopus drives (the Octopus Energy integration, for one) leaves the charger uncontrolled for good, and was only an Info line. It is now a warning naming the component's own setting to set true. Waiting for the Octopus component to discover its devices, which normally clears within a cycle or two, stays Info. Each reason is logged once, and again only when it changes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): wait for Kraken to wire its SmartFlex devices before driving a charger (#5311 review) Only the Octopus component was waited for. After a restart a Kraken (E.ON/EDF SmartFlex) user's charger could be driven before Kraken had wired octopus_intelligent_slot - stopping a GE charger mid-dispatch, which the hand-over on the next cycle then left stopped. Kraken wires its devices in its first successful run, from cache or fresh discovery, so until it has started the charger is left alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): release a held charger when Predbat cannot tell who drives it, and stop waiting on discovery after 15 minutes (#5311 review) - Cannot tell whether Octopus drives a charger: a charger Predbat had stopped was let go without being started, as if Octopus were taking it over. With nobody known to take it over, a GE charger was left stopped indefinitely. It is now released as usual; the non-starting hand-over is kept for a charger Octopus is known to drive. - Waiting for Octopus or Kraken discovery is bounded at 15 minutes. A Kraken component whose first run never succeeds (a failed login) otherwise left every unwired car's charger uncontrolled for good. The log line now names both components. - The apps.yaml default's literal "re:" string, before its regex matches, is no longer read as a dispatch sensor - which on every restart warned users with no Intelligent sensor to set a control override they do not need. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): leave a held charger as it is while discovery runs, and treat a discovery that never finishes as cannot tell (#5311 review) Releasing during discovery started a charge Predbat had stopped whenever the Octopus or Kraken component restarted, and driving the charger once the wait ran out could fight Kraken if it was the one driving it. While discovery is under way the charger is now left as it is and still held, so it is settled either way once discovery finishes. After 15 minutes Predbat still cannot tell, so it warns once and hands back a charger it holds, as for a dispatch sensor that does not say. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): start each wait for discovery afresh once a charger's driver is known (#5311 review) The wait's start was only cleared when discovery was asked about with nothing discovering, which stops happening once every car's driver is known. A later component restart could then find the old start and count the wait as already over. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(ohme): Predbat drives the Ohme only with ohme_control: true, so its own schedule stays the car plan otherwise Main's #5401 now takes the car plan from Ohme's own schedule when Predbat is not controlling the charger, so a default user already gets a plan that matches what the charger will do. Turning control on by default would override that schedule - one the user may have set up deliberately - with max charge, which also overrides the Ohme app's target. ohme_control goes back to a false default; GE and Zappi keep their automatic defaults. The size and limit guard for an unset ohme_control goes, as there is no unset default any more. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(car): octopus_intelligent_charger_follows_car - leave a charger alone when Octopus drives the car, unless asked to follow it When the car itself is the Octopus Intelligent device, the car starts and stops its own charge. Predbat now leaves the charger alone by default, handing over a charger it held without starting a charge, as it does when Octopus drives the charger. The new expert-mode switch, default off, has Predbat drive the charger to the car's dispatches instead, so the car cannot also charge on its own timers. The Ohme follows the same switch: with it on, ohme_control can run while Octopus drives the car, where #5405 otherwise stands it down. A change of the switch is picked up on the next Ohme poll. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): release a held charger when Octopus drives the car, keep following Predbat's own plan with Intelligent charging off, and log only real Ohme changes (#5311 review) - Octopus driving the car means nobody drives the charger, so a charger Predbat had stopped is released as usual (GE started, Zappi back to its own mode including Fast) rather than handed over without starting, which left it stopped through every dispatch. - With switch.predbat_octopus_intelligent_charging off Predbat plans the car itself, so it still drives the charger to that plan whatever the new switch says. - An Ohme poll that sees only the new switch change logs that change, and only re-decides control while Octopus drives another device - not the Octopus tariff change it used to report. - The log points at expert mode for the switch, and the class docstring says the charger is left alone by default. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(ohme): drive the Ohme to Predbat's own plan with Intelligent charging off while Octopus drives the car (#5311 review) The shared charger rule drives a charger for a car Octopus drives when switch.predbat_octopus_intelligent_charging is off, as Predbat then plans the car itself; the Ohme stood control down there regardless. It now follows the same rule, and re-decides when either switch changes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): read a dispatch time with no offset as UTC, and a lost is_charger value as cannot tell (local review) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * fix(car): hand an Ohme to Octopus without starting a charge, and warn when Storage refuses to save the control switch (#5448 review) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * refactor(car): share the hand-over to Octopus, and only cancel a max charge Predbat started (#5448 review) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * fix(car): time each car's wait for Octopus discovery on its own (local review) The discovery wait was one clock shared by every car, and drive_one reset it whenever any car's driver was known. With two cars - one wired, one still waiting on a Kraken login that keeps failing - the wired car reset the clock every cycle, so the waiting car never reached "cannot tell" and its charger was held as it was indefinitely, with no warning. The clock is now kept per car, and still cleared for every car once discovery has finished. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(debug-journal): restore control_active for the Sunsynk/Deye persistence clause (#5448 review) The myenergi rename to charger_control_active caught the Sunsynk/Deye sentence too; those components still persist control_active. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): leave a charger alone while its own car's plan is unpublished (#5448 review) charger_control_refresh_windows() only waited for any car's plan, so with car 0's slot sensor published and car 1's not, car 1 read as an empty plan and its charger was stopped. A car missing from charger_control_windows is now unknown: its charger is left as it is until its plan appears or an Octopus dispatch says to charge. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): keep driving a held charger whose car's plan is unpublished (local review) The unknown-plan skip came before the drift and last-state checks, so a charger a dispatch had started for a car with no published plan was frozen on once the dispatch ended. The skip now applies only to a charger Predbat does not hold. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): keep a car's last plan when its slot sensor briefly loses it (local review) A held charger whose car's planned attribute went missing for a cycle was driven from an empty window list and stopped mid-charge. The windows are absolute times, so the last ones read are carried forward instead. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(ohme): a refused hand-back no longer fails the whole poll (local review) update_slot_mode() handed the charger back outside any handler, so an ApiException from Ohme skipped the session fetch, publish and slot wiring for every poll until Ohme accepted the command. It is now a warning, as in the control tick; the charger stays held and the next poll retries. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(myenergi): save a Zappi's mode only once Predbat's command has gone through (local review) The mode was snapshotted ahead of set_mode, so a refused command left a snapshot for a Zappi Predbat never held. Release paths only run for held Zappis, so it could outlive the moment and be restored much later in place of what the user had set by then. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(output): publish each car's slot sensor with only its own windows publish_car_plan() built one plan list for every car, and handed the same list object to each slot sensor, so with two or more cars every sensor's 'planned' attribute held the windows of all of them. Pre-existing on main, but the shared charger control drives charger N from car N's 'planned', so a charger would start in another car's window. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * Revert "fix(myenergi): save a Zappi's mode only once Predbat's command has gone through (local review)" This reverts commit 3ff61b7. Snapshotting after set_mode traded one hazard for a worse one: a command that timed out but was applied is retried with the Zappi already Stopped, and Stopped was then saved as the user's mode, so a later release left the car unable to charge. Saving ahead of the command gets the mode the user really had; a snapshot left by a refused command is the lesser risk, and a full release still clears it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(myenergi): never release or hand over a Zappi back to Stopped (local review) A saved Stopped is most likely Predbat's own hold: after a restart, or a Stop that timed out but was applied, the first snapshot is taken with the Zappi already Stopped. Restoring it left the car unable to charge, the very outcome a release exists to prevent. It now falls back to Eco+, as for no saved mode. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(ohme): use charger_control_active in the #5443 dispatch-sensor tests The tests main added in #5443 read OhmeAPI.control_active, which this branch renamed to charger_control_active when control moved into the shared CarChargerControl mixin. Only the attribute name changes; the assertions are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(charger-control): judge each car on its own dispatch sensor under an Ohme claim Since #5443 the Octopus component keeps wiring its whole dispatch-sensor list into octopus_intelligent_slot while Ohme holds car_slot_owner (the claim now covers only the ready time and charge limit). The car_slot_owner shortcut in charger_control_octopus_drives_charger()/_drives_car() predates that and applied to every car, so with an Ohme on car 0 and a second car whose Intelligent device is the car itself, a Zappi or GE EVC on car 1 was left to Octopus: follows_car ignored, and a held charger handed over stopped. The sensor's own is_charger now decides; the claim only stands in when the sensor does not say (the Ohme's own slot sensor). drives_car no longer reads an "unknown" is_charger as "the car". Regression test fails on the merge without this change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(charger-control): keep the Ohme claim deciding car 0, as before the merge The previous commit let car 0's own dispatch sensor override an Ohme car_slot_owner claim. That changed behaviour the merge never touched: with ohme_automatic_octopus_intelligent forced on, or Octopus's device ids sorting the car's sensor first, a charger on car 0 that used to be left to Octopus would be released or driven. Car 0 goes back to the claim alone; only cars after it, which had no slot at all under the claim before #5443, are judged on their own sensor - and one whose sensor has lost its value is "cannot tell" rather than handed to Octopus. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(charger-control): say what an Ohme claim covers since #5443 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(car): charger control stays off unless set, and a Zappi's own mode survives a restart Review changes from Trefor on #5448. ge_cloud_evc_control and myenergi_zappi_control default to false again, so an upgrade never starts Predbat driving a charger nobody asked it to drive. Control runs only with an explicit true, and still stands down where Octopus Intelligent drives the charger or the car. MyEnergiAPI.automatic_set_by_user() is removed. The mode a Zappi had before Predbat took over is now saved to storage and restored on startup, so a restart no longer snapshots Predbat's own Stopped as the mode to release to. With that fixed, a saved Stopped is restored like any other mode, on release and on the hand-over to Octopus (which still never uses Fast). Modes left in storage are dropped when control is off at startup. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: CI <ci@example.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #5402. Builds on #5401, which is merged.
The problem
With
ohme_automatic: trueandohme_automatic_octopus_intelligentunset, the Ohme component took the Intelligent car slots for itself whenever the import tariff was an Intelligent one. It never checked which device Octopus Intelligent controls.When that device is the car (a BMW, Mini or Volkswagen linked to Octopus directly) and not the Ohme, Octopus schedules the charge through the car and the Ohme is only the socket. Predbat was then planning from Ohme's session, target and ready time, and ignoring Octopus's own dispatch list. The issue has an example: Octopus reporting no dispatches and 80% by 18:40, while Predbat worked from a 12-hour, 91 kWh Ohme slot to 100% by 11:52.
What this changes
Auto-detection now looks at the Octopus component's Intelligent devices as well as the tariff. On an Intelligent tariff with the flag unset:
ohme_controlstands downAn explicit
ohme_automatic_octopus_intelligent: trueorfalsestill wins, and the Ohme is still registered as the charger (car_charging_planned,car_charging_now,car_charging_energy) in every case.Decisions worth a look
is_chargertrue and providerOhme(case-insensitive), as the issue proposed. A charger of another make on the account is treated like the car: Octopus keeps the slots.OctopusAPI.automatic_config()already does for its own wiring. This does change behaviour for an Ohme whose smart charging is suspended in the Octopus app: its slots used to be priced as cheap dispatches and are now car load at the tariff rate. That seems right, since Octopus is not dispatching it, but it is wider than the issue's example.ohme_controlis ruled out when Octopus drives another device, not only when the slots come from Ohme. Otherwise Predbat would pause and max-charge the Ohme while Octopus is scheduling the car through it. With an explicitohme_automatic_octopus_intelligent: falseit is still allowed, as before.How it works
octopus_intelligent_wanted()reads the devices throughOctopusAPI.get_intelligent_devices(). That list only holds live devices and is kept through a failed poll, so it does not flap with the API. A newoctopus_other_deviceflag records "Octopus is driving something that is not this charger", andcharger_slots_wanted()refuses the charger-schedule mode from feat(ohme): take the car plan from Ohme's own schedule when Predbat is not controlling the charger #5401 while it is set.OctopusAPI.run()now re-wires the car slots when their owner changes. If Ohme claimed the slots before the device was known and then gives the claim up, the device set has not moved, so nothing else would have wired them. This is the one change outsideohme.py.Testing
test_ohme.py: the detection rules for each device combination (Ohme, car, other charger, both, suspended, none, explicit flag, non-Intelligent tariff); a first run with the car as the device, with and without Octopus's wiring already in place, and withohme_control; and a running install where the device is found late, swapped for the Ohme, then suspended.test_octopus_intelligent_devices.py: the Octopus component leaves the slots alone while Ohme holds them, wires its own entity on the next sensor refresh after Ohme lets go, and does not repeat that../run_all --quickpasses.Docs:
docs/car-charging.md(the auto-detect description and the "Which car charging plan Predbat shows" table) and the Ohme entry indocs/components.md.🤖 Generated with Claude Code