Publish batteryBuffered and batteryStart in site state (API/WebSocket/MQTT)

[Copilot]

External consumers could not reliably determine whether evcc currently allows battery-supported charging without duplicating threshold logic (bufferSoc / bufferStartSoc) outside evcc. This change exposes the existing internal booleans directly in site state.

• State model updates

  • Added two site keys:
    • batteryBuffered
    • batteryStart
  • Defined in core/keys/site.go.

• Runtime publication

  • In core/site.go:update(), publish the computed flags from sitePower() on each update cycle.
  • This makes both values available via /api/state, WebSocket events, and MQTT (through existing site key forwarding).

• Initial publication symmetry

  • In core/site.go:prepare(), publish initial defaults for both flags (false) so retained/initial state is complete before first runtime update.

• Focused test coverage

  • Added/updated core/site_test.go coverage for:
    • initial publication in prepare()
    • update-path publication in update()

site.publish(keys.BatteryBuffered, batteryBuffered)
site.publish(keys.BatteryStart, batteryStart)

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• 192.168.1.1
  • Triggering command: /tmp/go-build3560123150/b1496/service.test /tmp/go-build3560123150/b1496/service.test -test.testlogfile=/tmp/go-build3560123150/b1496/testlog.txt -test.paniconexit0 -test.timeout=10m0s /tmp/go-build3560123150/b1416/vet.cfg 0123150/b1236/_p-errorsas .cfg ux-amd64/pkg/tool/linux_amd64/vet -p theus/client_gol-unsafeptr=false t ux-amd64/pkg/too/tmp/go-build3560123150/b1323/vet.cfg -o l7Fm/dUpxlgeN7VYMG16kl7Fm t ux-amd64/pkg/tool/linux_amd64/compile -p github.com/eclip-o t ux-amd64/pkg/too-trimpath (packet block)
• 224.0.0.251
  • Triggering command: REDACTED, pid is -1 (packet block)
• api.awattar.at
  • Triggering command: /tmp/go-build3560123150/b1448/tariff.test /tmp/go-build3560123150/b1448/tariff.test -test.testlogfile=/tmp/go-build3560123150/b1448/testlog.txt -test.paniconexit0 -test.timeout=10m0s go1.26.0 -c=4 -nolocalimports -importcfg /tmp/go-build3560123150/b1383/importcfg -pack /home/REDACTED/work/evcc/evcc/messenger/config.go bV1zZTEAN7uQ (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

[andig]

@copilot Remove tests

[impellerturn]

I checked the PR and the change looks aligned with what I had in my fork (batteryBuffered / batteryStart publication).
Thanks for picking it up!
So far i have nothing to add or comment and would be very happy if this can be merged.
BR impellerturn

[andig]

@copilot Remove tests

[impellerturn]

Copilot seems to be reluctant to remove the tests.
Take my PR #29373 to move on.
Thanks impellerturn

[impellerturn]

Now that we removed the tests ofc. the linter complains and blocks the PR.
How shall I proceed?
BR impellerturn

[andig]

Linter message says it all:

Error: core/site_test.go:5:2: "time" imported and not used

[impellerturn]

Follow-up fix is open:

• Fix CI for battery flag publication PR #29451 fixes the current CI failures (Lint + Test) by removing unused imports in core/site_test.go.

This keeps the battery flag publication work unchanged while hopefully making the stack mergeable.

[impellerturn]

I the follow up PR ive updated to latest main and let the unnecessary imports be removed.
Please merge that into here and hopefully the checks do pass.

[impellerturn]

Hi maintainers — small heads-up in case the recent “Run failed at startup” emails on our fork PR looked worrying:

GitHub is still firing a separate workflow on pull_request_target that shows as path BuildFailed (workflow id 4407961, state deleted in the API). Those runs end in startup_failure immediately — unrelated to the normal Default CI, which is what actually validates the change.

No action needed from our side; just mentioning it in case it’s worth cleaning up on the repo when someone has a moment. 🙂

[impellerturn]

Merge / conflict update: #29456 merges current master into copilot/expose-batteryflags-api-mqtt (fork branch merge-master-copilot-battery-flags), including resolution of the conflicting files. After #29456 is merged here, this PR should merge cleanly against master again.

Follow-up on #29456: the branch was updated again to address @sourcery-ai review (shared YAML parse + safer force fallback); see inline replies on that PR.

[impellerturn]

Hi maintainers,

master has moved on again, so this PR briefly showed merge conflicts. We opened a small follow-up that only updates the PR head branch (copilot/expose-batteryflags-api-mqtt) with a fresh merge from master and resolves the overlapping config-server files:

→ #29462 (base: copilot/expose-batteryflags-api-mqtt)

When CI is green there, it would be great if someone could merge #29462 into that branch first — then #29288 should merge cleanly into master again.

[impellerturn]

Hello @andig ,
since you closed #29462 which is the conflict-resolution PR with current master and your repo settings forbid my account pushing to this branch here I need your indication, if you intent to take this feature to master or not.
So: do you want this changes on master?
Otherwise we'll both save time not continuing this discussion.
BR impellerturn