Add daemon-attached goal recovery operator

Resume goals after transient API failures with persisted exponential backoff while respecting pauses and interruptions.

codex-keep/store.go

@@ -0,0 +1,78 @@
+package main
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"os"
+	"path/filepath"
+	"syscall"
+)
+
+type store struct {
+	path    string
+	lock    *os.File
+	Watches map[string]*watch `json:"watches"`
+}
+
+func openStore(path string) (*store, error) {
+	if err := os.MkdirAll(filepath.Dir(path), 0700); err != nil {
+		return nil, err
+	}
+	lock, err := os.OpenFile(path+".lock", os.O_CREATE|os.O_RDWR, 0600)
+	if err != nil {
+		return nil, err
+	}
+	if err := syscall.Flock(int(lock.Fd()), syscall.LOCK_EX|syscall.LOCK_NB); err != nil {
+		return nil, fmt.Errorf("checkpoint is already in use: %w", errors.Join(err, lock.Close()))
+	}
+	s := &store{path: path, lock: lock, Watches: make(map[string]*watch)}
+	data, err := os.ReadFile(path)
+	if errors.Is(err, os.ErrNotExist) {
+		return s, nil
+	}
+	if err == nil {
+		err = json.Unmarshal(data, s)
+	}
+	if err == nil && s.Watches == nil {
+		err = errors.New("checkpoint has no watches")
+	}
+	if err != nil {
+		return nil, errors.Join(err, s.close())
+	}
+	return s, nil
+}
+
+func (s *store) close() error { return s.lock.Close() }
+
+func (s *store) save() error {
+	data, err := json.Marshal(s)
+	if err != nil {
+		return err
+	}
+	f, err := os.CreateTemp(filepath.Dir(s.path), ".state-*")
+	if err != nil {
+		return err
+	}
+	defer func() { _ = os.Remove(f.Name()) }()
+	if _, err = f.Write(data); err == nil {
+		err = f.Sync()
+	}
+	closeErr := f.Close()
+	if err != nil {
+		return err
+	}
+	if closeErr != nil {
+		return closeErr
+	}
+	return os.Rename(f.Name(), s.path)
+}
+
+func (s *store) watch(id string) *watch {
+	w := s.Watches[id]
+	if w == nil {
+		w = &watch{}
+		s.Watches[id] = w
+	}
+	return w
+}