Commit Graph

5 Commits

Author SHA1 Message Date
26a2aabbc7 leisure Phase 2 (dormant): bill, warn and kill on the leisure budget from go-live
Dormant until earned_time.leisure_active(gaming day) -- the switch file
written by `leisure-watch go-live`. From that day:

- enforced `seconds` are billed by the shadow rule (lock pause, fail-closed,
  once per wall clock); budget from plain resolve(); warnings and the cutoff
  say "Leisure".
- at 0, beside the unchanged Steam path, _leisure_kill SIGTERMs (SIGKILL 30 s
  later, per process) every focus/media pid in now.json except browsers,
  every failclosed browser pid, and -- with now.json untrusted -- every
  browser, media player and wsg-grabber reviewer. Root/system, lockers and
  the enforcer's own chain are never signalled. Kills go to budget.jsonl.
- tests: an autouse fixture redirects both switch readers, so no test reads
  /etc/leisure-watch/leisure_from; existing tests unchanged.

Re-pins earned-time to v0.15.0 (leisure_active).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dsh4Znwh2KnC7q6vbGpX1N
2026-10-19 16:40:49 +02:00
b5a76d6b45 shadow: forget expired warnings so varying messages cannot grow memory
The once-per-window warning map kept every distinct error forever; an
error whose text varies per tick would grow the daemon without bound.
Entries older than the window are now dropped before each check.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dsh4Znwh2KnC7q6vbGpX1N
2026-10-19 13:20:08 +02:00
f4f480c1ac shadow: no exception can escape into the enforce loop; warn once per pricing
observe_shadow caught only OSError/ValueError, so a RecursionError from
leisure_watch escaped into the unguarded root loop and crash-looped the
daemon, stopping enforced billing. It now catches Exception, fails closed
(failclosed:watcher, unless the pause check itself answered "paused") and
logs each distinct error at most once per 10 minutes.

resolve_budget priced the same answers twice (enforced + leisure), so
earned_time's "No answer for the X earner" warning fired twice per tick.
The shadow pricing now gets the answers completed with None, which
earned_time treats identically to missing, so only the first warns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dsh4Znwh2KnC7q6vbGpX1N
2026-10-19 13:16:16 +02:00
79e5e9fc26 shadow: fail closed instead of raising on unreadable lock/fact inputs
Some checks are pending
dependency freshness / dependencies on newest stable (push) Waiting to run
md-naming / markdown naming convention (push) Waiting to run
pre-commit / pre-commit (${{ matrix.group }}) (mypy, mypy, ) (push) Waiting to run
pre-commit / pre-commit (${{ matrix.group }}) (pylint, pylint, ) (push) Waiting to run
pre-commit / pre-commit (${{ matrix.group }}) (rest, , mypy,pylint) (push) Waiting to run
Tests / test (push) Waiting to run
Tests / web-coverage (push) Waiting to run
repo-contract / repo contract (push) Waiting to run
The enforce loop has no per-tick guard, so an OSError/ValueError from
paused_by or read_now would crash-loop the daemon and stop enforced
billing. Such a tick now bills shadow time unpaused (Q8) and logs a warning.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dsh4Znwh2KnC7q6vbGpX1N
2026-10-19 12:50:41 +02:00
e82ddd9a4b playtime: shadow leisure billing (C4); /proc scan from leisure_watch
The enforcer now imports the Steam/launcher/counted-process scan from
leisure_watch.procscan (v0.1.0, shared with the leisure watcher) and drops
its own copies (_playtime_procs, _counted_procs, _total_block_launchers and
enforcer's two scanners). Every counted-process read passes
config.CONFIG_FILE explicitly, as before.

Each tick also bills a shadow leisure budget beside the enforced one, from
the same incoming stamp and delta clamp, before accumulate: paused (held
gatelock grab or i3lock) -> shadow_paused_seconds; else a qualifying game,
a "leisure" now.json, or an untrusted/missing now.json (fail closed) bills
shadow_seconds once, split over shadow_per_source. shadow_budget_seconds is
earned_time.resolve(leisure=True) from the same answers, held as a same-day
high-water mark. Not billed under a total block. Shadow verdict changes go
to budget.jsonl as shadow_change (the live game view skips them); /api/budget
and MCP get_gaming_time carry a "shadow" block. Enforced fields, cutoff,
warnings and existing log records are unchanged.

Also fixes two tests that assumed earned_time < 0.14 internals.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dsh4Znwh2KnC7q6vbGpX1N
2026-10-19 12:37:01 +02:00