opensource/kubo
1
0
Fork 0
mirror of https://github.com/ipfs/kubo.git synced 2025-06-28 17:03:58 +08:00
Code Issues Projects Releases Wiki Activity

Merge pull request #925 from jbenet/fsrepo-cleanup

Browse Source 
Fsrepo cleanup
This commit is contained in:
Juan Batiz-Benet
2015-03-15 23:41:16 -07:00
parent bc55fb539f 03bc2ccb9d
commit bd357d934e
16 changed files with 217 additions and 314 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
cmd/ipfs/daemon.go
View File

@ -118,8 +118,8 @@ func daemonFunc(req cmds.Request, res cmds.Response) {
// acquire the repo lock _before_ constructing a node. we need to make
// sure we are permitted to access the resources (datastore, etc.)
repo := fsrepo.At(req.Context().ConfigRoot)
if err := repo.Open(); err != nil {
repo, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(debugerror.Errorf("Couldn't obtain lock. Is another daemon already running?"), cmds.ErrNormal)
return
}

8
cmd/ipfs/init.go
View File

@ -110,8 +110,8 @@ func addDefaultAssets(out io.Writer, repoRoot string) error {
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
r := fsrepo.At(repoRoot)
if err := r.Open(); err != nil { // NB: repo is owned by the node
r, err := fsrepo.Open(repoRoot)
if err != nil { // NB: repo is owned by the node
return err
}
@ -163,8 +163,8 @@ func initializeIpnsKeyspace(repoRoot string) error {
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
r := fsrepo.At(repoRoot)
if err := r.Open(); err != nil { // NB: repo is owned by the node
r, err := fsrepo.Open(repoRoot)
if err != nil { // NB: repo is owned by the node
return err
}

4
cmd/ipfs/main.go
View File

@ -193,8 +193,8 @@ func (i *cmdInvocation) constructNodeFunc(ctx context.Context) func() (*core.Ipf
return nil, errors.New("constructing node without a request context")
}
r := fsrepo.At(i.req.Context().ConfigRoot)
if err := r.Open(); err != nil { // repo is owned by the node
r, err := fsrepo.Open(i.req.Context().ConfigRoot)
if err != nil { // repo is owned by the node
return nil, err
}

4
cmd/ipfs/tour.go
View File

@ -193,8 +193,8 @@ func tourGet(id tour.ID) (*tour.Topic, error) {
// TODO share func
func writeConfig(path string, cfg *config.Config) error {
// NB: This needs to run on the daemon.
r := fsrepo.At(path)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(path)
if err != nil {
return err
}
defer r.Close()

4
cmd/ipfs_bootstrapd/main.go
View File

@ -57,8 +57,8 @@ func run() error {
}
}
repo := fsrepo.At(repoPath)
if err := repo.Open(); err != nil { // owned by node
repo, err := fsrepo.Open(repoPath)
if err != nil { // owned by node
return err
}

4
cmd/ipfs_routingd/main.go
View File

@ -57,8 +57,8 @@ func run() error {
return err
}
}
repo := fsrepo.At(repoPath)
if err := repo.Open(); err != nil { // owned by node
repo, err := fsrepo.Open(repoPath)
if err != nil { // owned by node
return err
}

4
cmd/ipfswatch/main.go
View File

@ -65,8 +65,8 @@ func run(ipfsPath, watchPath string) error {
return err
}
r := fsrepo.At(ipfsPath)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(ipfsPath)
if err != nil {
// TODO handle case: daemon running
// TODO handle case: repo doesn't exist or isn't initialized
return err

12
core/commands/bootstrap.go
View File

@ -66,8 +66,8 @@ in the bootstrap list).
return
}
r := fsrepo.At(req.Context().ConfigRoot)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(err, cmds.ErrNormal)
return
}
@ -143,8 +143,8 @@ var bootstrapRemoveCmd = &cmds.Command{
return
}
r := fsrepo.At(req.Context().ConfigRoot)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(err, cmds.ErrNormal)
return
}
@ -192,8 +192,8 @@ var bootstrapListCmd = &cmds.Command{
},
Run: func(req cmds.Request, res cmds.Response) {
r := fsrepo.At(req.Context().ConfigRoot)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(err, cmds.ErrNormal)
return
}

9
core/commands/config.go
View File

@ -64,14 +64,13 @@ Set the value of the 'datastore.path' key:
args := req.Arguments()
key := args[0]
r := fsrepo.At(req.Context().ConfigRoot)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(err, cmds.ErrNormal)
return
}
defer r.Close()
var err error
var output *ConfigField
if len(args) == 2 {
value := args[1]
@ -182,8 +181,8 @@ can't be undone.
cmds.FileArg("file", true, false, "The file to use as the new config"),
},
Run: func(req cmds.Request, res cmds.Response) {
r := fsrepo.At(req.Context().ConfigRoot)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(req.Context().ConfigRoot)
if err != nil {
res.SetError(err, cmds.ErrNormal)
return
}

43
repo/fsrepo/counter/openers.go
View File

@ -1,43 +0,0 @@
package counter
import "path"
// TODO this could be made into something more generic.
type Openers struct {
// repos maps repo paths to the number of openers holding an FSRepo handle
// to it
repos map[string]int
}
func NewOpenersCounter() *Openers {
return &Openers{
repos: make(map[string]int),
}
}
// NumOpeners returns the number of FSRepos holding a handle to the repo at
// this path. This method is not thread-safe. The caller must have this object
// locked.
func (l *Openers) NumOpeners(repoPath string) int {
return l.repos[key(repoPath)]
}
// AddOpener messages that an FSRepo holds a handle to the repo at this path.
// This method is not thread-safe. The caller must have this object locked.
func (l *Openers) AddOpener(repoPath string) error {
l.repos[key(repoPath)]++
return nil
}
// RemoveOpener messgaes that an FSRepo no longer holds a handle to the repo at
// this path. This method is not thread-safe. The caller must have this object
// locked.
func (l *Openers) RemoveOpener(repoPath string) error {
l.repos[key(repoPath)]--
return nil
}
func key(repoPath string) string {
return path.Clean(repoPath)
}

263
repo/fsrepo/fsrepo.go
View File

@ -1,8 +1,6 @@
package fsrepo
import (
"errors"
"fmt"
"io"
"os"
"path"
@ -15,7 +13,6 @@ import (
repo "github.com/jbenet/go-ipfs/repo"
"github.com/jbenet/go-ipfs/repo/common"
config "github.com/jbenet/go-ipfs/repo/config"
counter "github.com/jbenet/go-ipfs/repo/fsrepo/counter"
lockfile "github.com/jbenet/go-ipfs/repo/fsrepo/lock"
serialize "github.com/jbenet/go-ipfs/repo/fsrepo/serialize"
dir "github.com/jbenet/go-ipfs/thirdparty/dir"
@ -34,41 +31,33 @@ var (
// packageLock must be held to while performing any operation that modifies an
// FSRepo's state field. This includes Init, Open, Close, and Remove.
packageLock sync.Mutex // protects openersCounter and lockfiles
// lockfiles holds references to the Closers that ensure that repos are
// only accessed by one process at a time.
lockfiles map[string]io.Closer
// openersCounter prevents the fsrepo from being removed while there exist open
// FSRepo handles. It also ensures that the Init is atomic.
//
// packageLock also protects numOpenedRepos
//
// If an operation is used when repo is Open and the operation does not
// change the repo's state, the package lock does not need to be acquired.
openersCounter *counter.Openers
packageLock sync.Mutex
// protects dsOpenersCounter and datastores
dsLock sync.Mutex
dsOpenersCounter *counter.Openers
datastores map[string]ds2.ThreadSafeDatastoreCloser
// onlyOne keeps track of open FSRepo instances.
//
// TODO: once command Context / Repo integration is cleaned up,
// this can be removed. Right now, this makes ConfigCmd.Run
// function try to open the repo twice:
//
// $ ipfs daemon &
// $ ipfs config foo
//
// The reason for the above is that in standalone mode without the
// daemon, `ipfs config` tries to save work by not building the
// full IpfsNode, but accessing the Repo directly.
onlyOne repo.OnlyOne
)
func init() {
openersCounter = counter.NewOpenersCounter()
lockfiles = make(map[string]io.Closer)
dsOpenersCounter = counter.NewOpenersCounter()
datastores = make(map[string]ds2.ThreadSafeDatastoreCloser)
}
// FSRepo represents an IPFS FileSystem Repo. It is safe for use by multiple
// callers.
type FSRepo struct {
// state is the FSRepo's state (unopened, opened, closed)
state state
// has Close been called already
closed bool
// path is the file-system path
path string
// config is set on Open, guarded by packageLock
// lockfile is the file system lock to prevent others from opening
// the same fsrepo path concurrently
lockfile io.Closer
config *config.Config
// ds is set on Open
ds ds2.ThreadSafeDatastoreCloser
@ -76,13 +65,63 @@ type FSRepo struct {
var _ repo.Repo = (*FSRepo)(nil)
// At returns a handle to an FSRepo at the provided |path|.
func At(repoPath string) *FSRepo {
// This method must not have side-effects.
return &FSRepo{
path: path.Clean(repoPath),
state: unopened, // explicitly set for clarity
// Open the FSRepo at path. Returns an error if the repo is not
// initialized.
func Open(repoPath string) (repo.Repo, error) {
fn := func() (repo.Repo, error) {
return open(repoPath)
}
return onlyOne.Open(repoPath, fn)
}
func open(repoPath string) (repo.Repo, error) {
packageLock.Lock()
defer packageLock.Unlock()
expPath, err := u.TildeExpansion(path.Clean(repoPath))
if err != nil {
return nil, err
}
r := &FSRepo{
path: expPath,
}
r.lockfile, err = lockfile.Lock(r.path)
if err != nil {
return nil, err
}
keepLocked := false
defer func() {
// unlock on error, leave it locked on success
if !keepLocked {
r.lockfile.Close()
}
}()
if !isInitializedUnsynced(r.path) {
return nil, debugerror.New("ipfs not initialized, please run 'ipfs init'")
}
// check repo path, then check all constituent parts.
// TODO acquire repo lock
// TODO if err := initCheckDir(logpath); err != nil { // }
if err := dir.Writable(r.path); err != nil {
return nil, err
}
if err := r.openConfig(); err != nil {
return nil, err
}
if err := r.openDatastore(); err != nil {
return nil, err
}
// log.Debugf("writing eventlogs to ...", c.path)
configureEventLoggerAtRepoPath(r.config, r.path)
keepLocked = true
return r, nil
}
// ConfigAt returns an error if the FSRepo at the given path is not
@ -165,15 +204,6 @@ func Init(repoPath string, conf *config.Config) error {
// Remove recursively removes the FSRepo at |path|.
func Remove(repoPath string) error {
repoPath = path.Clean(repoPath)
// packageLock must be held to ensure that the repo is not removed while
// being accessed by others.
packageLock.Lock()
defer packageLock.Unlock()
if openersCounter.NumOpeners(repoPath) != 0 {
return errors.New("repo in use")
}
return os.RemoveAll(repoPath)
}
@ -182,12 +212,11 @@ func Remove(repoPath string) error {
func LockedByOtherProcess(repoPath string) bool {
repoPath = path.Clean(repoPath)
// packageLock must be held to check the number of openers.
packageLock.Lock()
defer packageLock.Unlock()
// TODO replace this with the "api" file
// https://github.com/ipfs/specs/tree/master/repo/fs-repo
// NB: the lock is only held when repos are Open
return lockfile.Locked(repoPath) && openersCounter.NumOpeners(repoPath) == 0
return lockfile.Locked(repoPath)
}
// openConfig returns an error if the config file is not present.
@ -206,32 +235,14 @@ func (r *FSRepo) openConfig() error {
// openDatastore returns an error if the config file is not present.
func (r *FSRepo) openDatastore() error {
dsLock.Lock()
defer dsLock.Unlock()
dsPath := path.Join(r.path, defaultDataStoreDirectory)
// if no other goroutines have the datastore Open, initialize it and assign
// it to the package-scoped map for the goroutines that follow.
if dsOpenersCounter.NumOpeners(dsPath) == 0 {
ds, err := levelds.NewDatastore(dsPath, &levelds.Options{
Compression: ldbopts.NoCompression,
})
if err != nil {
return debugerror.New("unable to open leveldb datastore")
}
datastores[dsPath] = ds
}
// get the datastore from the package-scoped map and record self as an
// opener.
ds, dsIsPresent := datastores[dsPath]
if !dsIsPresent {
// This indicates a programmer error has occurred.
return errors.New("datastore should be available, but it isn't")
}
r.ds = ds
dsOpenersCounter.AddOpener(dsPath) // only after success
return nil
}
@ -247,76 +258,16 @@ func configureEventLoggerAtRepoPath(c *config.Config, repoPath string) {
eventlog.Configure(eventlog.OutputRotatingLogFile(rotateConf))
}
// Open returns an error if the repo is not initialized.
func (r *FSRepo) Open() error {
// packageLock must be held to make sure that the repo is not destroyed by
// another caller. It must not be released until initialization is complete
// and the number of openers is incremeneted.
packageLock.Lock()
defer packageLock.Unlock()
expPath, err := u.TildeExpansion(r.path)
if err != nil {
return err
}
r.path = expPath
if r.state != unopened {
return debugerror.Errorf("repo is %s", r.state)
}
if !isInitializedUnsynced(r.path) {
return debugerror.New("ipfs not initialized, please run 'ipfs init'")
}
// check repo path, then check all constituent parts.
// TODO acquire repo lock
// TODO if err := initCheckDir(logpath); err != nil { // }
if err := dir.Writable(r.path); err != nil {
return err
}
if err := r.openConfig(); err != nil {
return err
}
if err := r.openDatastore(); err != nil {
return err
}
// log.Debugf("writing eventlogs to ...", c.path)
configureEventLoggerAtRepoPath(r.config, r.path)
return r.transitionToOpened()
}
func (r *FSRepo) closeDatastore() error {
dsLock.Lock()
defer dsLock.Unlock()
dsPath := path.Join(r.path, defaultDataStoreDirectory)
// decrement the Opener count. if this goroutine is the last, also close
// the underlying datastore (and remove its reference from the map)
dsOpenersCounter.RemoveOpener(dsPath)
if dsOpenersCounter.NumOpeners(dsPath) == 0 {
delete(datastores, dsPath) // remove the reference
return r.ds.Close()
}
return nil
}
// Close closes the FSRepo, releasing held resources.
func (r *FSRepo) Close() error {
packageLock.Lock()
defer packageLock.Unlock()
if r.state != opened {
return debugerror.Errorf("repo is %s", r.state)
if r.closed {
return debugerror.New("repo is closed")
}
if err := r.closeDatastore(); err != nil {
if err := r.ds.Close(); err != nil {
return err
}
@ -328,7 +279,11 @@ func (r *FSRepo) Close() error {
// to disable logging once the component is closed.
// eventlog.Configure(eventlog.Output(os.Stderr))
return r.transitionToClosed()
r.closed = true
if err := r.lockfile.Close(); err != nil {
return err
}
return nil
}
// Config returns the FSRepo's config. This method must not be called if the
@ -345,8 +300,8 @@ func (r *FSRepo) Config() *config.Config {
packageLock.Lock()
defer packageLock.Unlock()
if r.state != opened {
panic(fmt.Sprintln("repo is", r.state))
if r.closed {
panic("repo is closed")
}
return r.config
}
@ -393,8 +348,8 @@ func (r *FSRepo) GetConfigKey(key string) (interface{}, error) {
packageLock.Lock()
defer packageLock.Unlock()
if r.state != opened {
return nil, debugerror.Errorf("repo is %s", r.state)
if r.closed {
return nil, debugerror.New("repo is closed")
}
filename, err := config.Filename(r.path)
@ -413,8 +368,8 @@ func (r *FSRepo) SetConfigKey(key string, value interface{}) error {
packageLock.Lock()
defer packageLock.Unlock()
if r.state != opened {
return debugerror.Errorf("repo is %s", r.state)
if r.closed {
return debugerror.New("repo is closed")
}
filename, err := config.Filename(r.path)
@ -479,37 +434,3 @@ func isInitializedUnsynced(repoPath string) bool {
}
return true
}
// transitionToOpened manages the state transition to |opened|. Caller must hold
// the package mutex.
func (r *FSRepo) transitionToOpened() error {
r.state = opened
if countBefore := openersCounter.NumOpeners(r.path); countBefore == 0 { // #first
closer, err := lockfile.Lock(r.path)
if err != nil {
return err
}
lockfiles[r.path] = closer
}
return openersCounter.AddOpener(r.path)
}
// transitionToClosed manages the state transition to |closed|. Caller must
// hold the package mutex.
func (r *FSRepo) transitionToClosed() error {
r.state = closed
if err := openersCounter.RemoveOpener(r.path); err != nil {
return err
}
if countAfter := openersCounter.NumOpeners(r.path); countAfter == 0 {
closer, ok := lockfiles[r.path]
if !ok {
return errors.New("package error: lockfile is not held")
}
if err := closer.Close(); err != nil {
return err
}
delete(lockfiles, r.path)
}
return nil
}

58
repo/fsrepo/fsrepo_test.go
View File

@ -30,31 +30,7 @@ func TestInitIdempotence(t *testing.T) {
func TestRemove(t *testing.T) {
t.Parallel()
path := testRepoPath("foo", t)
assert.Nil(Remove(path), t, "should be able to remove after closed")
}
func TestCannotRemoveIfOpen(t *testing.T) {
t.Parallel()
path := testRepoPath("TestCannotRemoveIfOpen", t)
assert.Nil(Init(path, &config.Config{}), t, "should initialize successfully")
r := At(path)
assert.Nil(r.Open(), t)
assert.Err(Remove(path), t, "should not be able to remove while open")
assert.Nil(r.Close(), t)
assert.Nil(Remove(path), t, "should be able to remove after closed")
}
func TestCannotBeReopened(t *testing.T) {
t.Parallel()
path := testRepoPath("", t)
assert.Nil(Init(path, &config.Config{}), t)
r := At(path)
assert.Nil(r.Open(), t)
assert.Nil(r.Close(), t)
assert.Err(r.Open(), t, "shouldn't be possible to re-open the repo")
// mutable state is the enemy. Take Close() as an opportunity to reduce
// entropy. Callers ought to start fresh with a new handle by calling `At`.
assert.Nil(Remove(path), t, "can remove a repository")
}
func TestCanManageReposIndependently(t *testing.T) {
@ -71,10 +47,10 @@ func TestCanManageReposIndependently(t *testing.T) {
assert.True(IsInitialized(pathB), t, "b should be initialized")
t.Log("open the two repos")
repoA := At(pathA)
repoB := At(pathB)
assert.Nil(repoA.Open(), t, "a")
assert.Nil(repoB.Open(), t, "b")
repoA, err := Open(pathA)
assert.Nil(err, t, "a")
repoB, err := Open(pathB)
assert.Nil(err, t, "b")
t.Log("close and remove b while a is open")
assert.Nil(repoB.Close(), t, "close b")
@ -91,15 +67,15 @@ func TestDatastoreGetNotAllowedAfterClose(t *testing.T) {
assert.True(!IsInitialized(path), t, "should NOT be initialized")
assert.Nil(Init(path, &config.Config{}), t, "should initialize successfully")
r := At(path)
assert.Nil(r.Open(), t, "should open successfully")
r, err := Open(path)
assert.Nil(err, t, "should open successfully")
k := "key"
data := []byte(k)
assert.Nil(r.Datastore().Put(datastore.NewKey(k), data), t, "Put should be successful")
assert.Nil(r.Close(), t)
_, err := r.Datastore().Get(datastore.NewKey(k))
_, err = r.Datastore().Get(datastore.NewKey(k))
assert.Err(err, t, "after closer, Get should be fail")
}
@ -108,16 +84,16 @@ func TestDatastorePersistsFromRepoToRepo(t *testing.T) {
path := testRepoPath("test", t)
assert.Nil(Init(path, &config.Config{}), t)
r1 := At(path)
assert.Nil(r1.Open(), t)
r1, err := Open(path)
assert.Nil(err, t)
k := "key"
expected := []byte(k)
assert.Nil(r1.Datastore().Put(datastore.NewKey(k), expected), t, "using first repo, Put should be successful")
assert.Nil(r1.Close(), t)
r2 := At(path)
assert.Nil(r2.Open(), t)
r2, err := Open(path)
assert.Nil(err, t)
v, err := r2.Datastore().Get(datastore.NewKey(k))
assert.Nil(err, t, "using second repo, Get should be successful")
actual, ok := v.([]byte)
@ -131,11 +107,11 @@ func TestOpenMoreThanOnceInSameProcess(t *testing.T) {
path := testRepoPath("", t)
assert.Nil(Init(path, &config.Config{}), t)
r1 := At(path)
r2 := At(path)
assert.Nil(r1.Open(), t, "first repo should open successfully")
assert.Nil(r2.Open(), t, "second repo should open successfully")
assert.True(r1.ds == r2.ds, t, "repos should share the datastore")
r1, err := Open(path)
assert.Nil(err, t, "first repo should open successfully")
r2, err := Open(path)
assert.Nil(err, t, "second repo should open successfully")
assert.True(r1 == r2, t, "second open returns same value")
assert.Nil(r1.Close(), t)
assert.Nil(r2.Close(), t)

22
repo/fsrepo/state.go
View File

@ -1,22 +0,0 @@
package fsrepo
type state int
const (
unopened = iota
opened
closed
)
func (s state) String() string {
switch s {
case unopened:
return "unopened"
case opened:
return "opened"
case closed:
return "closed"
default:
return "invalid"
}
}

72
repo/onlyone.go Normal file
View File

@ -0,0 +1,72 @@
package repo
import (
"sync"
)
// OnlyOne tracks open Repos by arbitrary key and returns the already
// open one.
type OnlyOne struct {
mu sync.Mutex
active map[interface{}]*ref
}
// Open a Repo identified by key. If Repo is not already open, the
// open function is called, and the result is remember for further
// use.
//
// Key must be comparable, or Open will panic. Make sure to pick keys
// that are unique across different concrete Repo implementations,
// e.g. by creating a local type:
//
// type repoKey string
// r, err := o.Open(repoKey(path), open)
//
// Call Repo.Close when done.
func (o *OnlyOne) Open(key interface{}, open func() (Repo, error)) (Repo, error) {
o.mu.Lock()
defer o.mu.Unlock()
if o.active == nil {
o.active = make(map[interface{}]*ref)
}
item, found := o.active[key]
if !found {
repo, err := open()
if err != nil {
return nil, err
}
item = &ref{
parent: o,
key: key,
Repo: repo,
}
o.active[key] = item
}
item.refs++
return item, nil
}
type ref struct {
parent *OnlyOne
key interface{}
refs uint32
Repo
}
var _ Repo = (*ref)(nil)
func (r *ref) Close() error {
r.parent.mu.Lock()
defer r.parent.mu.Unlock()
r.refs--
if r.refs > 0 {
// others are holding it open
return nil
}
// last one
delete(r.parent.active, r.key)
return r.Repo.Close()
}

4
test/supernode_client/main.go
View File

@ -66,8 +66,8 @@ func run() error {
repoPath := gopath.Join(cwd, ".go-ipfs")
if err := ensureRepoInitialized(repoPath); err != nil {
}
repo := fsrepo.At(repoPath)
if err := repo.Open(); err != nil { // owned by node
repo, err := fsrepo.Open(repoPath)
if err != nil { // owned by node
return err
}
cfg := repo.Config()

4
updates/updates.go
View File

@ -212,8 +212,8 @@ func CliCheckForUpdates(cfg *config.Config, repoPath string) error {
// if we checked successfully.
if err == ErrNoUpdateAvailable {
log.Noticef("No update available, checked on %s", time.Now())
r := fsrepo.At(repoPath)
if err := r.Open(); err != nil {
r, err := fsrepo.Open(repoPath)
if err != nil {
return err
}
if err := recordUpdateCheck(cfg); err != nil {