Documentation
¶
Overview ¶
Package storage provides a filesystem based storage layer for tony system api.
Storage manages
mappings of virtual document paths to and from the filesyste
indexing of virtual document nodes
storage of diffs associated with virtual document nodes
transactions of multi-participant diffs
compaction of diffs
Index ¶
- type Storage
- func (s *Storage) Close() error
- func (s *Storage) GetCurrentCommit() (int64, error)
- func (s *Storage) GetTx(txID int64) (tx.Tx, error)
- func (s *Storage) NewTx(participantCount int, meta *api.PatchMeta) (tx.Tx, error)
- func (s *Storage) ReadStateAt(kp string, commit int64) (*ir.Node, error)
- func (s *Storage) SwitchAndSnapshot() error
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Storage ¶
type Storage struct {
// contains filtered or unexported fields
}
Storage provides filesystem-based storage for logd.
func Open ¶
Open opens or creates a Storage instance with the given root directory. The root directory will be created if it doesn't exist. umask is applied to directory permissions (e.g., 022 for 0755 -> 0755). If logger is nil, slog.Default() will be used. compactorConfig is optional - if nil, a default config with divisor 2 and NeverRemove is used.
func (*Storage) GetCurrentCommit ¶ added in v0.0.10
GetCurrentCommit returns the current commit number.
func (*Storage) GetTx ¶ added in v0.0.10
GetTx gets an existing transaction by transaction ID. This is the primary way participants coordinate - they all receive the same transaction ID and get the same transaction.
Example:
// Multiple parallel HTTP handlers all receive the same txID
tx, err := storage.GetTx(txID)
if err != nil {
// handle error
}
// Each participant gets their own patcher handle
patcher := tx.NewPatcher(kp, m, p)
result := patcher.Commit()
func (*Storage) NewTx ¶ added in v0.0.10
NewTx creates a new transaction with the specified number of participants. Returns a transaction that participants can get via GetTx or get a patcher via NewPatcher().
Example usage (typical pattern for parallel HTTP handlers):
// Create transaction
tx, err := storage.NewTx(participantCount, meta)
if err != nil {
// handle error
}
// Each participant gets their own patcher handle
patcher := tx.NewPatcher(kp, m, p)
result := patcher.WaitForCompletion()
func (*Storage) ReadStateAt ¶ added in v0.0.10
ReadStateAt reads the state for a given kpath at a specific commit count. Searches for the most recent snapshot and applies patches from that point forward.
func (*Storage) SwitchAndSnapshot ¶ added in v0.0.15
SwitchAndSnapshot switches the active log and creates a snapshot of the inactive log. The snapshot is created for the current commit at the time of switching. This should be called periodically (e.g., based on log size or time) to enable snapshot-based read optimization and eventual compaction. Protected by switchMu to prevent concurrent switching while snapshot is being written.
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
Package index provides hierarchical path-based indexing for storage.
|
Package index provides hierarchical path-based indexing for storage. |
|
internal
|
|
|
dlog
Package dlog provides double-buffered write-ahead logging.
|
Package dlog provides double-buffered write-ahead logging. |
|
patches
Package patches provides streaming patch application for snapshots.
|
Package patches provides streaming patch application for snapshots. |
|
seq
Package seq provides atomic sequence counters.
|
Package seq provides atomic sequence counters. |
|
snap
Package snap provides event-based snapshot storage.
|
Package snap provides event-based snapshot storage. |
|
Package tx provides transaction coordination for multi-participant updates.
|
Package tx provides transaction coordination for multi-participant updates. |