ltx

package module
v0.5.3 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Sep 10, 2026 License: Apache-2.0 Imports: 20 Imported by: 25

README

Lite Transaction File (LTX)

The LTX file format provides a way to store SQLite transactional data in a way that can be encrypted and compacted and is optimized for performance.

File Format

This document describes format version 3. LTX files carry no on-disk version field, and versions 2 and 3 both use the LTX1 magic number. Version 2 page frames used a four-byte page header with no flags or compressed-size prefix. Version 3 uses a six-byte page header and, in the current encoding, a four-byte compressed-size prefix. A reader cannot determine the format version from the file alone and must know it out of band.

An LTX file is composed of four sections:

  1. Header
  2. Page block
  3. Page index
  4. Trailer

The header contains metadata about the file, the page block contains page frames, the page index enables random access to frames, and the trailer contains checksums for the file and the database end state. Unless otherwise specified, all fixed-width integer fields use big-endian byte order.

Header

The 100-byte header describes the database state and transaction ID (TXID) range represented by the file.

Offset Size Field Description
0 4 Magic Always LTX1.
4 4 Flags Header flags.
8 4 PageSize Database page size, in bytes.
12 4 Commit Database size after applying the file, in pages.
16 8 MinTXID Minimum transaction ID.
24 8 MaxTXID Maximum transaction ID.
32 8 Timestamp Milliseconds since the Unix epoch.
40 8 PreApplyChecksum Database checksum before applying the file.
48 8 WALOffset Offset in the original WAL; zero for a journal.
56 8 WALSize WAL segment size; zero for a journal.
64 4 WALSalt1 First WAL salt; zero for a journal or compaction.
68 4 WALSalt2 Second WAL salt; zero for journal or compaction.
72 8 NodeID Creator node ID; zero if unset.
80 20 Reserved Written as zero by the current encoder.
Header flags
Flag Name Description
0x00000002 HeaderFlagNoChecksum Disable database checksum tracking.

HeaderFlagNoChecksum is bit 1 (1 << 1). When set, the pre-apply and post-apply database checksums are zero. All other header flag bits are currently invalid. The file checksum is still required when database checksum tracking is disabled.

Page block

The page block stores page frames in ascending page-number order. Each frame written by the current encoder has this layout:

Offset Size Field Description
0 4 Pgno One-based database page number.
4 2 Flags Page header flags.
6 4 CompressedSize Size of the compressed payload, in bytes.
10 N Data LZ4 block-compressed database page data.
Page header flags
Flag Name Description
0x0001 PageHeaderFlagSize A four-byte compressed-size field follows.

PageHeaderFlagSize is bit 0 (1 << 0). The payload must decompress to Header.PageSize bytes. The current encoder always sets this flag. Within version 3, the decoder uses it as a per-frame encoding heuristic and supports legacy frames without the flag; those store page data as an LZ4 frame without a size prefix. The flag is not a format version field. All other page header flag bits are invalid.

A six-byte zero page header terminates the page block and has no size prefix or page data.

Page index

The page index follows the zero page header. It contains one entry per page, sorted by page number. Each entry is encoded as three consecutive unsigned varints:

  1. Page number
  2. Absolute byte offset of the page frame from the start of the file
  3. Encoded frame size, including its header, optional size prefix, and payload

A zero page-number varint terminates the entries. An eight-byte big-endian unsigned integer follows and contains the total byte size of the varint entries, including the zero terminator but excluding the size field itself.

Trailer

The 16-byte trailer contains the database checksum after applying the file and the file checksum.

Offset Size Field Description
0 8 PostApplyChecksum Database checksum after applying the file.
8 8 FileChecksum CRC-ISO-64 checksum described below.
File checksum

The file checksum is a CRC-ISO-64 over the following input, in order:

  1. The header bytes.
  2. For every page, the page header and compressed-size prefix, when present, as stored, followed by the decompressed page data instead of the compressed payload bytes.
  3. The zero page header that terminates the page block.
  4. All page index bytes, including its zero terminator and size field.
  5. The trailer's post-apply checksum field.

The final eight-byte file checksum field is not included. The stored value is the calculated CRC with ChecksumFlag (1 << 63) set.

This checksum provides logical-content integrity plus integrity for selected structural metadata; it is not a byte-for-byte checksum of the file on disk. In particular, different valid LZ4 payload bytes produce the same checksum when they decompress to the same page data and do not change the hashed size or index values.

Documentation

Overview

Package ltx reads and writes Liteserver Transaction (LTX) files.

Index

Constants

View Source 
const (
	// Magic is the first 4 bytes of every LTX file.
	Magic = "LTX1"

	// Version is the current version of the LTX file format.
	Version = 3
)
View Source 
const (
	HeaderSize     = 100
	PageHeaderSize = 6
	TrailerSize    = 16
)

Size constants.

View Source 
const (
	ChecksumSize          = 8
	TrailerChecksumOffset = TrailerSize - ChecksumSize
)

Checksum size & positions.

View Source 
const (
	HeaderFlagMask = uint32(HeaderFlagNoChecksum)

	HeaderFlagNoChecksum = uint32(1 << 1)
)

Header flags.

View Source 
const DefaultSpillThreshold = 1 << 20

DefaultSpillThreshold is the number of page index entries an Encoder keeps in memory before spilling the index to disk, when a spill directory is set. 1M entries is ~24 MiB in memory; a 4 KiB-page database crosses it at 4 GiB.

View Source 
const MaxPageSize = 65536

MaxPageSize is the maximum allowed page size for SQLite.

View Source 
const PENDING_BYTE = 0x40000000
View Source 
const (
	// PageHeaderFlagSize indicates that a 4-byte size field follows the page
	// header. When set, data uses LZ4 block format (not frame format).
	PageHeaderFlagSize = uint16(1 << 0)
)

PageHeader flags.

View Source 
const RFC3339Milli = "2006-01-02T15:04:05.000Z07:00"

RFC3339Milli is the standard time format for LTX timestamps. It uses fixed-width millisecond resolution which makes it sortable.

Variables

View Source 
var (
	ErrInvalidFile   = errors.New("invalid LTX file")
	ErrDecoderClosed = errors.New("ltx decoder closed")
	ErrEncoderClosed = errors.New("ltx encoder closed")

	// ErrEncoderAborted is returned by an Encoder whose file cannot be
	// completed: Cleanup ran before a successful Close, or an earlier write
	// or Close failed.
	ErrEncoderAborted = errors.New("ltx encoder aborted")

	ErrNoChecksum            = errors.New("no file checksum")
	ErrInvalidChecksumFormat = errors.New("invalid file checksum format")
	ErrChecksumMismatch      = errors.New("file checksum mismatch")
)

Errors

Functions

func ChecksumPages added in v0.3.14 

func ChecksumPages(dbPath string, pageSize, nPages, nWorkers uint32, checksums []Checksum) (uint32, error)

ChecksumPages updates the provided checksums slice with the checksum of each page in the specified file. The first (by page number) error encountered is returned along with the number of the last page successfully checksummed. Checksums for subsequent pages may be updated, regardless of an error being returned.

nWorkers specifies the amount of parallelism to use. A reasonable default will be used if nWorkers is 0.

func DecodePageIndex added in v0.4.0 

func DecodePageIndex(r io.ByteReader, level int, minTXID, maxTXID TXID) (map[uint32]PageIndexElem, error)

DecodePageIndex decodes the page index from r. It validates that page numbers ascend, that offsets and sizes are in range, and that the trailing size field matches the bytes read; it cannot check the index against the page block, so callers reading only the tail of a file must treat the result as untrusted range metadata.

func FormatFilename 

func FormatFilename(minTXID, maxTXID TXID) string

FormatFilename returns an LTX filename representing a range of transactions.

func FormatTimestamp added in v0.3.12 

func FormatTimestamp(t time.Time) string

FormatTimestamp returns t with a fixed-width, millisecond-resolution UTC format.

func IsContiguous added in v0.3.17 

func IsContiguous(prevMaxTXID, minTXID, maxTXID TXID) bool

IsContiguous returns true if the transaction range is contiguous with the previous range. This is used to validate that the transaction ranges are contiguous when rebuilding snapshots.

func IsValidHeaderFlags 

func IsValidHeaderFlags(flags uint32) bool

IsValidHeaderFlags returns true unless flags outside the valid mask are set.

func IsValidPageSize 

func IsValidPageSize(sz uint32) bool

IsValidPageSize returns true if sz is between 512 and 64K and a power of two.

func LockPgno added in v0.2.11 

func LockPgno(pageSize uint32) uint32

LockPgno returns the page number where the PENDING_BYTE exists.

func NewHasher added in v0.2.10 

func NewHasher() hash.Hash64

NewHasher returns a new CRC64-ISO hasher.

func ParseTimestamp added in v0.3.12 

func ParseTimestamp(value string) (time.Time, error)

ParseTimestamp parses a timestamp as RFC3339Milli (fixed-width) but will fallback to RFC3339Nano if it fails. This is to support timestamps written before the introduction of the standard time format.

Types

type Checksum added in v0.3.13 

type Checksum uint64

Checksum represents an LTX checksum. 

const ChecksumFlag Checksum = 1 << 63

ChecksumFlag is a flag on the checksum to ensure it is non-zero.

func ChecksumPage 

func ChecksumPage(pgno uint32, data []byte) Checksum

ChecksumPage returns a CRC64 checksum that combines the page number & page data.

func ChecksumPageWithHasher added in v0.2.10 

func ChecksumPageWithHasher(h hash.Hash64, pgno uint32, data []byte) Checksum

ChecksumPageWithHasher returns a CRC64 checksum that combines the page number & page data.

func ChecksumReader added in v0.2.1 

func ChecksumReader(r io.Reader, pageSize int) (Checksum, error)

ChecksumReader reads an entire database file from r and computes its rolling checksum.

func ParseChecksum added in v0.3.13 

func ParseChecksum(s string) (Checksum, error)

ParseChecksum parses a 16-character hex string into a checksum.

func (Checksum) MarshalJSON added in v0.3.13 

func (c Checksum) MarshalJSON() ([]byte, error)

func (Checksum) String added in v0.3.13 

func (c Checksum) String() string

String returns c formatted as a fixed-width hex number.

func (*Checksum) UnmarshalJSON added in v0.3.13 

func (c *Checksum) UnmarshalJSON(data []byte) (err error)

type Compactor 

type Compactor struct {

	// These flags will be set when encoding the header.
	HeaderFlags uint32

	// If true, the compactor will not validate that input files have contiguous
	// transaction IDs. This is false by default but can be enabled when
	// rebuilding snapshots with missing transactions.
	AllowNonContiguousTXIDs bool
	// contains filtered or unexported fields
}

Compactor represents a compactor of LTX files.

func NewCompactor 

func NewCompactor(w io.Writer, rdrs []io.Reader) (*Compactor, error)

NewCompactor returns a new instance of Compactor with default settings.

func (*Compactor) Cleanup added in v0.5.3 

func (c *Compactor) Cleanup() error

Cleanup removes any spill file left by an abandoned compaction and aborts the output encoder if it was not closed successfully. It is safe to call after Compact returns, successfully or not. See Encoder.Cleanup.

func (*Compactor) Compact 

func (c *Compactor) Compact(ctx context.Context) (err error)

Compact merges the input readers into a single LTX writer. A failed compaction cannot be resumed, so its output spill file, if any, is removed before returning.

func (*Compactor) Header added in v0.3.3 

func (c *Compactor) Header() Header

Header returns the LTX header of the compacted file. Only valid after successful Compact().

func (*Compactor) SetSpillDir added in v0.5.3 

func (c *Compactor) SetSpillDir(dir string)

SetSpillDir enables spilling the output page index to a temp file in dir once it grows past the encoder's spill threshold. See Encoder.SetSpillDir.

func (*Compactor) SetSpillThreshold added in v0.5.3 

func (c *Compactor) SetSpillThreshold(n int)

SetSpillThreshold sets the number of in-memory page index entries that triggers a spill of the output index. See Encoder.SetSpillThreshold.

func (*Compactor) Status added in v0.5.1 

func (c *Compactor) Status() CompactorStatus

Status returns the current compaction progress. Safe to call concurrently from another goroutine.

func (*Compactor) Trailer added in v0.3.3 

func (c *Compactor) Trailer() Trailer

Trailer returns the LTX trailer of the compacted file. Only valid after successful Compact().

type CompactorStatus added in v0.5.1 

type CompactorStatus struct {
	N     uint32 // Last page number that was compacted
	Total uint32 // Total pages to compact (from Commit field)
}

CompactorStatus represents the current progress of compaction.

func (CompactorStatus) IsZero added in v0.5.1 

func (s CompactorStatus) IsZero() bool

IsZero returns true if the status has not been initialized.

func (CompactorStatus) Pct added in v0.5.1 

func (s CompactorStatus) Pct() float64

Pct returns the percentage of compaction complete (0.0 to 1.0). Returns 0 if Total is zero.

type Decoder added in v0.2.2 

type Decoder struct {
	// contains filtered or unexported fields
}

Decoder represents a decoder of an LTX file.

func NewDecoder added in v0.2.2 

func NewDecoder(r io.Reader) *Decoder

NewDecoder returns a new instance of Decoder.

func (*Decoder) Close added in v0.2.2 

func (dec *Decoder) Close() error

Close verifies the reader is at the end of the file and that the checksum matches.

func (*Decoder) DecodeDatabaseTo added in v0.3.4 

func (dec *Decoder) DecodeDatabaseTo(w io.Writer) error

DecodeDatabaseTo decodes the LTX file as a SQLite database to w. The LTX file MUST be a snapshot file.

func (*Decoder) DecodeHeader added in v0.2.2 

func (dec *Decoder) DecodeHeader() error

DecodeHeader reads the LTX file header frame and stores it internally. Call Header() to retrieve the header after this is successfully called.

func (*Decoder) DecodePage added in v0.2.2 

func (dec *Decoder) DecodePage(hdr *PageHeader, data []byte) error

DecodePage reads the next page header into hdr and associated page data.

func (*Decoder) Header added in v0.2.2 

func (dec *Decoder) Header() Header

Header returns a copy of the header.

func (*Decoder) N added in v0.2.2 

func (dec *Decoder) N() int64

N returns the number of bytes read.

func (*Decoder) PageIndex added in v0.4.0 

func (dec *Decoder) PageIndex() map[uint32]PageIndexElem

PageIndex returns a mapping of page numbers to byte offsets and sizes of those pages. This returns the raw reference and not a copy.

func (*Decoder) PageN added in v0.3.0 

func (dec *Decoder) PageN() int

PageN returns the number of pages read.

func (*Decoder) PostApplyPos added in v0.3.6 

func (dec *Decoder) PostApplyPos() Pos

PostApplyPos returns the replication position after underlying the LTX file is applied. Only valid after successful Close().

func (*Decoder) SetRetainPageIndex added in v0.5.3 

func (dec *Decoder) SetRetainPageIndex(retain bool)

SetRetainPageIndex controls whether Close builds the in-memory page index returned by PageIndex. It defaults to true. When false, Close still reads, validates, and checksums the index but discards the entries, so memory no longer scales with the number of pages in the file; PageIndex returns nil. Must be called before Close.

func (*Decoder) Trailer added in v0.2.2 

func (dec *Decoder) Trailer() Trailer

Trailer returns a copy of the trailer. File checksum available after Close().

func (*Decoder) Verify added in v0.2.2 

func (dec *Decoder) Verify() error

Verify reads the entire file. Header & trailer can be accessed via methods after the file is successfully verified. All other data is discarded.

type Encoder added in v0.2.2 

type Encoder struct {
	// contains filtered or unexported fields
}

Encoder implements an encoder for an LTX file.

func NewEncoder added in v0.2.2 

func NewEncoder(w io.Writer) (*Encoder, error)

NewEncoder returns a new instance of Encoder.

func (*Encoder) Cleanup added in v0.5.3 

func (enc *Encoder) Cleanup() error

Cleanup removes the spill file, if any, and aborts an encoder that has not completed a successful Close, so a later Close cannot emit a file whose index was discarded. It is idempotent, safe after Close, and must be called by callers that abandon an encoder without closing it (for example when a snapshot upload is cancelled) so the spill file is not left behind. On removal failure the spill is kept so Cleanup can be retried; the error is returned.

func (*Encoder) Close added in v0.2.2 

func (enc *Encoder) Close() (err error)

Close flushes the checksum to the header.

func (*Encoder) EncodeHeader added in v0.2.2 

func (enc *Encoder) EncodeHeader(hdr Header) error

EncodeHeader writes hdr to the file's header block.

func (*Encoder) EncodePage added in v0.2.2 

func (enc *Encoder) EncodePage(hdr PageHeader, data []byte) (err error)

EncodePage writes hdr & data to the file's page block.

func (*Encoder) Header added in v0.2.2 

func (enc *Encoder) Header() Header

Header returns a copy of the header.

func (*Encoder) N added in v0.2.2 

func (enc *Encoder) N() int64

N returns the number of bytes written.

func (*Encoder) PostApplyPos added in v0.3.6 

func (enc *Encoder) PostApplyPos() Pos

PostApplyPos returns the replication position after underlying the LTX file is applied. Only valid after successful Close().

func (*Encoder) SetPostApplyChecksum added in v0.2.2 

func (enc *Encoder) SetPostApplyChecksum(chksum Checksum)

SetPostApplyChecksum sets the post-apply checksum of the database. Must call before Close().

func (*Encoder) SetSpillDir added in v0.5.3 

func (enc *Encoder) SetSpillDir(dir string)

SetSpillDir enables spilling the page index to a temp file created in dir once it reaches the spill threshold. The directory must be writable for the lifetime of the encoder; litestream passes its per-database meta directory since hardened images have no /tmp. Empty (the default) keeps the whole index in memory.

func (*Encoder) SetSpillThreshold added in v0.5.3 

func (enc *Encoder) SetSpillThreshold(n int)

SetSpillThreshold sets the number of in-memory page index entries that triggers a spill. Values <= 0 restore DefaultSpillThreshold.

func (*Encoder) Spilled added in v0.5.3 

func (enc *Encoder) Spilled() bool

Spilled reports whether the page index has been spilled to disk.

func (*Encoder) Trailer added in v0.2.2 

func (enc *Encoder) Trailer() Trailer

Trailer returns a copy of the trailer. File checksum available after Close().

type FileInfo added in v0.3.15 

type FileInfo struct {
	Level             int
	MinTXID           TXID
	MaxTXID           TXID
	PreApplyChecksum  Checksum
	PostApplyChecksum Checksum
	Size              int64
	CreatedAt         time.Time
}

FileInfo represents file information about an LTX file.

func SliceFileIterator added in v0.3.15 

func SliceFileIterator(itr FileIterator) ([]*FileInfo, error)

SliceFileIterator returns all LTX files from an iterator as a slice.

func (*FileInfo) Pos added in v0.3.15 

func (info *FileInfo) Pos() Pos

PostApplyPos returns the replication position after the LTX file is applied.

func (*FileInfo) PreApplyPos added in v0.3.15 

func (info *FileInfo) PreApplyPos() Pos

PreApplyPos returns the replication position before the LTX file is applied.

type FileInfoSlice added in v0.3.15 

type FileInfoSlice []*FileInfo

FileInfoSlice represents a slice of WAL segment metadata.

func (FileInfoSlice) Len added in v0.3.15 

func (a FileInfoSlice) Len() int

func (FileInfoSlice) Less added in v0.3.15 

func (a FileInfoSlice) Less(i, j int) bool

func (FileInfoSlice) MaxTXID added in v0.3.16 

func (a FileInfoSlice) MaxTXID() TXID

MaxTXID returns the MaxTXID of the last element in the slice. Returns zero if the slice is empty.

func (FileInfoSlice) Swap added in v0.3.15 

func (a FileInfoSlice) Swap(i, j int)

type FileInfoSliceIterator added in v0.3.15 

type FileInfoSliceIterator struct {
	// contains filtered or unexported fields
}

FileInfoSliceIterator represents an iterator for iterating over a slice of LTX files.

func NewFileInfoSliceIterator added in v0.3.15 

func NewFileInfoSliceIterator(a []*FileInfo) *FileInfoSliceIterator

NewFileInfoSliceIterator returns a new instance of FileInfoSliceIterator. This function will sort the slice in place before returning the iterator.

func (*FileInfoSliceIterator) Close added in v0.3.15 

func (itr *FileInfoSliceIterator) Close() error

Close always returns nil.

func (*FileInfoSliceIterator) Err added in v0.3.15 

func (itr *FileInfoSliceIterator) Err() error

Err always returns nil.

func (*FileInfoSliceIterator) Item added in v0.3.15 

func (itr *FileInfoSliceIterator) Item() *FileInfo

Item returns the metadata from the currently positioned wal segment.

func (*FileInfoSliceIterator) Next added in v0.3.15 

func (itr *FileInfoSliceIterator) Next() bool

Next moves to the next wal segment. Returns true if another segment is available.

type FileIterator added in v0.3.15 

type FileIterator interface {
	io.Closer

	// Prepares the next LTX file for reading with the Item() method.
	// Returns true if another item is available. Returns false if no more
	// items are available or if an error occured.
	Next() bool

	// Returns an error that occurred during iteration.
	Err() error

	// Returns metadata for the currently positioned LTX file.
	Item() *FileInfo
}

FileIterator represents an iterator over a collection of LTX files.

type FileSpec 

type FileSpec struct {
	Header  Header
	Pages   []PageSpec
	Trailer Trailer
}

FileSpec is an in-memory representation of an LTX file. Typically used for testing.

func (*FileSpec) GoString added in v0.3.1 

func (s *FileSpec) GoString() string

GoString returns the Go representation of s.

func (*FileSpec) ReadFrom 

func (s *FileSpec) ReadFrom(src io.Reader) (n int64, err error)

ReadFromFile encodes a file spec to a file. Always return n of zero.

func (*FileSpec) WriteTo 

func (s *FileSpec) WriteTo(dst io.Writer) (n int64, err error)

Write encodes a file spec to a file. 

type Header struct {
	Version          int      // based on magic
	Flags            uint32   // reserved flags
	PageSize         uint32   // page size, in bytes
	Commit           uint32   // db size after transaction, in pages
	MinTXID          TXID     // minimum transaction ID
	MaxTXID          TXID     // maximum transaction ID
	Timestamp        int64    // milliseconds since unix epoch
	PreApplyChecksum Checksum // rolling checksum of database before applying this LTX file
	WALOffset        int64    // file offset from original WAL; zero if journal
	WALSize          int64    // size of original WAL segment; zero if journal
	WALSalt1         uint32   // header salt-1 from original WAL; zero if journal or compaction
	WALSalt2         uint32   // header salt-2 from original WAL; zero if journal or compaction
	NodeID           uint64   // node id where the LTX file was created, zero if unset
}

Header represents the header frame of an LTX file.

func DecodeHeader added in v0.3.0 

func DecodeHeader(r io.Reader) (hdr Header, data []byte, err error)

DecodeHeader decodes the header from r. Returns the header & read bytes.

func PeekHeader added in v0.3.7 

func PeekHeader(r io.Reader) (Header, io.Reader, error)

PeekHeader reads & unmarshals the header from r. It returns a new io.Reader that prepends the header data back on.

func (*Header) IsSnapshot 

func (h *Header) IsSnapshot() bool

IsSnapshot returns true if header represents a complete database snapshot. This is true if the header includes the initial transaction. Snapshots must include all pages in the database.

func (*Header) LockPgno added in v0.3.11 

func (h *Header) LockPgno() uint32

LockPgno returns the lock page number based on the header's page size.

func (*Header) MarshalBinary 

func (h *Header) MarshalBinary() ([]byte, error)

MarshalBinary encodes h to a byte slice.

func (Header) NoChecksum added in v0.3.15 

func (h Header) NoChecksum() bool

NoChecksum returns true if the checksum is not being tracked for this LTX file.

func (Header) PreApplyPos added in v0.3.6 

func (h Header) PreApplyPos() Pos

PreApplyPos returns the replication position before the LTX file is applies.

func (*Header) UnmarshalBinary 

func (h *Header) UnmarshalBinary(b []byte) error

UnmarshalBinary decodes h from a byte slice.

func (*Header) Validate 

func (h *Header) Validate() error

Validate returns an error if h is invalid. 

type PageHeader struct {
	Pgno  uint32
	Flags uint16
}

PageHeader represents the header for a single page in an LTX file.

func DecodePageData added in v0.4.0 

func DecodePageData(b []byte) (hdr PageHeader, data []byte, err error)

DecodePageData decodes the page header & data from a single frame.

func (*PageHeader) IsZero added in v0.2.0 

func (h *PageHeader) IsZero() bool

IsZero returns true if the header is empty.

func (*PageHeader) MarshalBinary 

func (h *PageHeader) MarshalBinary() ([]byte, error)

MarshalBinary encodes h to a byte slice.

func (*PageHeader) UnmarshalBinary 

func (h *PageHeader) UnmarshalBinary(b []byte) error

UnmarshalBinary decodes h from a byte slice.

func (*PageHeader) Validate 

func (h *PageHeader) Validate() error

Validate returns an error if h is invalid.

type PageIndexElem added in v0.4.0 

type PageIndexElem struct {
	Level   int
	MinTXID TXID
	MaxTXID TXID

	Offset int64
	Size   int64
}

type PageSpec added in v0.2.0 

type PageSpec struct {
	Header PageHeader
	Data   []byte
}

PageSpec is an in-memory representation of an LTX page frame. Typically used for testing.

func (*PageSpec) GoString added in v0.3.1 

func (s *PageSpec) GoString() string

GoString returns the Go representation of s.

type Pos added in v0.3.1 

type Pos struct {
	TXID              TXID
	PostApplyChecksum Checksum
}

Pos represents the transactional position of a database.

func NewPos added in v0.3.3 

func NewPos(txID TXID, postApplyChecksum Checksum) Pos

NewPos returns a new instance of Pos.

func ParsePos added in v0.3.10 

func ParsePos(s string) (Pos, error)

ParsePos parses Pos from its string representation.

func (Pos) IsZero added in v0.3.1 

func (p Pos) IsZero() bool

IsZero returns true if the position is empty.

func (Pos) String added in v0.3.1 

func (p Pos) String() string

String returns a string representation of the position.

type PosMismatchError added in v0.3.1 

type PosMismatchError struct {
	Pos Pos `json:"pos"`
}

PosMismatchError is returned when an LTX file is not contiguous with the current position.

func NewPosMismatchError added in v0.3.1 

func NewPosMismatchError(pos Pos) *PosMismatchError

NewPosMismatchError returns a new instance of PosMismatchError.

func (*PosMismatchError) Error added in v0.3.1 

func (e *PosMismatchError) Error() string

Error returns the string representation of the error.

type TXID added in v0.3.2 

type TXID uint64

TXID represents a transaction ID.

func ParseFilename 

func ParseFilename(name string) (minTXID, maxTXID TXID, err error)

ParseFilename parses a transaction range from an LTX file.

func ParseTXID added in v0.2.2 

func ParseTXID(s string) (TXID, error)

ParseTXID parses a 16-character hex string into a transaction ID.

func (TXID) MarshalJSON added in v0.3.2 

func (t TXID) MarshalJSON() ([]byte, error)

func (TXID) String added in v0.3.2 

func (t TXID) String() string

String returns id formatted as a fixed-width hex number.

func (*TXID) UnmarshalJSON added in v0.3.2 

func (t *TXID) UnmarshalJSON(data []byte) (err error)

type Trailer added in v0.2.0 

type Trailer struct {
	PostApplyChecksum Checksum // rolling checksum of database after this LTX file is applied
	FileChecksum      Checksum // crc64 checksum of logical file content and structural metadata
}

Trailer represents the ending frame of an LTX file.

func (*Trailer) MarshalBinary added in v0.2.0 

func (t *Trailer) MarshalBinary() ([]byte, error)

MarshalBinary encodes h to a byte slice.

func (*Trailer) UnmarshalBinary added in v0.2.0 

func (t *Trailer) UnmarshalBinary(b []byte) error

UnmarshalBinary decodes h from a byte slice.

func (*Trailer) Validate added in v0.2.0 

func (t *Trailer) Validate(h Header) error

Validate returns an error if t is invalid.

Source Files

View all Source files

Directories

Path Synopsis
cmd
ltx command

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.