wfa

package module
v0.4.1 Latest Latest
Warning

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

 Go to latest Published: Jul 26, 2025 License: MIT Imports: 8 Imported by: 1

README

Wavefront alignment algorithm (WFA) in Golang

Go Reference

This golang package implements Wavefront alignment algorithm (WFA), not BiWFA (maybe in the future).

Implemented features:

  • Distance metrics: gap-affine.
  • Alignment types: global, semi-global.
  • Heuristics: wf-adaptive.

Executable binaries for common operating systems are available to download, for benchmarking with lots of sequences, or for fast aligning two sequences in the command line.

$ wfa-go -g "Bioinformatics helps Biology" "We learn bioinformatics to help biologists"
query   ---------Bioinformatics ---helps Biology---
                  ||||||||||||||   |||| | |||||
target  We learn bioinformatics to help- biologists
cigar   9I1X14M3I4M1D1M1X5M1X3I

align-score : 32
match-region: q[2, 27]/28 vs t[11, 38]/42
align-length: 29, matches: 24 (82.76%), gaps: 4, gap regions: 2

Table of Contents

History

  1. I need a fast DNA alignment package in Golang for my project.
  2. WFA "looks easy" to implement as it does not heavily reply on SIMD intrinsics, though there are some other algorithms performing well in a benchmark.
  3. I'm not familiar with C++, and I found it difficult to understand the official code. I only found one (in Rust) 3rd party implementation.
  4. After reading the WFA paper, I thought the algorithm is easy, so I implemented WFA from the scratch.
  5. Later I found it not easy, there were so many details. After reading the next step in the rust version, I realised it's because I don't know gap-affine penalties.
  6. The backtrace step is the most difficult part. In v0.1.0, I used 3 extra components to store the source offsets for each cell in I, D, M components. But it needs more memory. And the speed is also not ideal, 1/20 of official version.
    • Besides, I checked the bases again when tracing back matches. WFA did this too, but WFA2 did not.
  7. Next, aftering reading thofficial implementation, I rewrote the whole project, using similar backtrace workfow with WFA2. The speed increased to 1/10 of the official version.
  8. C++ is wild, it even support accessing list/array elements with negative indexes (the diagonal k).

Details

  • A WFA component is saved as a list of WaveFront []*WaveFront.

    • Score s is the index.
    • If the value is nil, it means the score does not exsit.

  • A WaveFront is saved as a list of offsets []uint32. We preset the list with a big length (2048) to avoid frequent append operations.

    • Diagonal k is the index. To support negative k values, we use ZigZag encoding:

      index: 0,  1,  2,  3,  4,  5,  6
k:     0, -1,  1, -2,  2, -3,  3

      Value 0 means the k does not exist.

    • Offsets are saved with uint32 integers, with the lower 3 bits for saving 6 possible paths which are used for backtrace.

      wfaInsertOpen uint32 = iota + 1
wfaInsertExt
wfaDeleteOpen
wfaDeleteExt
wfaMismatch
wfaMatch

  • Maximum sequence length: 512 Mb, 536,870,911 (1<<(32-3) - 1).

  • All objects are saved in object pools for computation efficiency. Just don't forget to recycle them.

Examples

Each WFA component can be visualized as a table. A table cell contains the alignment type symbol and the score.

⊕    Unknown,
⟼    Gap open (Insertion)
🠦    Gap extension (Insertion)
↧    Gap open (Deletion)
🠧    Gap extension (Deletion)
⬂    Mismatch
⬊    Match

Global alignment

1 2 3 4 5 6 7 8 9 10
A G G A T G C T C G
1 A ⬊ 0 ⟼ 8 🠦10 🠦12 . . . . . .
2 C ↧ 8 ⬂ 4 ⬂12 . . . . . . .
3 C 🠧10 ⬂12 ⬂ 8 . . . . . . .
4 A 🠧12 . . ⬊ 8 . . . . . .
5 T . . . . ⬊ 8 . . . . .
6 A . . . . . ⬂12 . . . .
7 C . . . . . . ⬊12 . . .
8 T . . . . . . . ⬊12 . .
9 C . . . . . . . . ⬊12 .
10 G . . . . . . . . . ⬊12 
CIGAR:  1M2X2M1X4M
query   ACCATACTCG
        |  || ||||
target  AGGATGCTCG

align-score : 12
align-region: q[1, 10] vs t[1, 10]
align-length: 10, matches: 7 (70.00%), gaps: 0, gapRegions: 0

Semi-global alignment

1 2 3 4 5 6 7 8 9 10 11 12
C A G G C T C C T C G G
1 A ⬂ 4 ⬊ 0 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4 ⬂ 4
2 C ⬊ 0 ⬂ 8 ⬂ 4 ⬂ 8 ⬊ 4 ⬂ 8 ⬊ 4 ⬊ 4 ⬂ 8 ⬊ 4 ⬂ 8 ⬂ 8
3 G ⬂ 4 ⬂ 4 ⬊ 8 ⬊ 4 ⬂12 ⬂ 8 ⬂12 ⬂ 8 ⬂ 8 . ⬊ 4 ⬊ 8
4 A ⬂ 4 ⬊ 4 ⬂ 8 ⬂12 ⬂ 8 🠧14 ⬂12 🠧14 ⬂12 ⬂12 ↧12 ⬂ 8
5 T ⬂ 4 ⬂ 8 ⬂ 8 ⬂12 . ⬊ 8 . ⬂16 ⬊14 . 🠧14 ⬂16
6 C ⬊ 0 ⬂ 8 🠦10 ⬂12 ⬊12 🠦18 ⬊ 8 ⟼16 🠦18 ⬊14 🠧16 ⬂18
7 T ⬂ 4 ⬂ 4 . . . ⬊12 ↧16 ⬂12 ⬊16 . ⬂18 ⬂20
8 C ⬊ 0 ⬂ 8 ⬂ 8 🠦12 🠦14 🠦16 ⬊12 ⬊16 ⬂16 ⬊16 🠧20 .
9 G ⬂ 4 ⬂ 4 ⬊ 8 ⬊ 8 ⬂16 ⬂18 ⬂20 ⬂16 ⬂20 ⬂20 ⬊16 ⬊20 
CIGAR:  1I1M1X1M1X1M1I4M1I
query   -ACGAT-CTCG-
         | | | ||||
target  CAGGCTCCTCGG

align-score : 16
align-region: q[1, 9] vs t[2, 11]
align-length: 10, matches: 7 (70.00%), gaps: 1, gapRegions: 1

Usages

import "github.com/shenwei356/wfa"

// ------------------[ initialization ]------------------

// aligner
algn := wfa.New(
    &wfa.Penalties{
        Mismatch: 4,
        GapOpen:  6,
        GapExt:   2,
    },
    &wfa.Options{
        GlobalAlignment: false,
    })

// set adaptive reduction parameters
algn.AdaptiveReduction(&wfa.AdaptiveReductionOption{
    MinWFLen:    10,
    MaxDistDiff: 50,
    CutoffStep:  1,
})

// ------------------[ align one pair of seqs ]------------------

q := []byte("ACCATACTCG")
t := []byte("AGGATGCTCG")

// align
result, err := algn.Align(q, t)
checkErr(err)

// score table of M
algn.Plot(&q, &t, os.Stdout, algn.M, true)

if outputAlignment {
    fmt.Println()
    fmt.Printf("CIGAR:  %s\n", result.CIGAR())

    Q, A, T := result.AlignmentText(&q, &t)
    fmt.Printf("query   %s\n", *Q)
    fmt.Printf("        %s\n", *A)
    fmt.Printf("target  %s\n", *T)

    fmt.Println()
    fmt.Printf("align-score : %d\n", result.Score)
    fmt.Printf("align-region: q[%d, %d] vs t[%d, %d]\n",
        result.QBegin, result.QEnd, result.TBegin, result.TEnd)
    fmt.Printf("align-length: %d, matches: %d (%.2f%%), gaps: %d, gapRegions: %d\n",
        result.AlignLen, result.Matches, float64(result.Matches)/float64(result.AlignLen)*100,
        result.Gaps, result.GapRegions)
    fmt.Println()

    
    wfa.RecycleAlignmentText(Q, A, T) // !! important, recycle objects
}

wfa.RecycleAlignmentResult(result) // !! important, recycle objects

// ------------------[ clean ]------------------

wfa.RecycleAligner(algn) // !! important, recycle objects

CLI

A CLI (download binaries) is available to align two sequences from either positional arguments or an input file (format).

Example

Fast alignment.

$ wfa-go AGCTAGTGTCAATGGCTACTTTTCAGGTCCT AACTAAGTGTCGGTGGCTACTATATATCAGGTCCT
query   AGCTA-GTGTCAATGGCTACT---TTTCAGGTCCT
        | ||| |||||  ||||||||   | |||||||||
target  AACTAAGTGTCGGTGGCTACTATATATCAGGTCCT
cigar   1M1X3M1I5M2X8M3I1M1X9M

align-score : 36
match-region: q[1, 31]/31 vs t[1, 35]/35
align-length: 35, matches: 27 (77.14%), gaps: 4, gap regions: 2

From a input file, for benchmark.

$ wfa-go -i seqs.txt
query   A-TTGGAAAATAGGATTGGGGTTTGTTTATATTTGGGTTGAGGGATGTCCCACCTTCGTCGTCCTTACGTTTCCGGAAGGGAGTGGTTAGCTCGAAGCCCA
          |||||||||||||| ||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||| ||||||||||||
target  GATTGGAAAATAGGAT-GGGGTTTGTTTATATTTGGGTTGAGGGATGTCCCACCTT-GTCGTCCTTACGTTTCCGGAAGGGAGTGGTT-GCTCGAAGCCCA
cigar   1X1I14M1D39M1D31M1D12M

align-score : 36
match-region: q[2, 100]/100 vs t[3, 98]/98
align-length: 99, matches: 96 (96.97%), gaps: 3, gap regions: 3
Usage 
WFA alignment in Golang

 Author: Wei Shen <shenwei356@gmail.com>
   Code: https://github.com/shenwei356/wfa
Version: v0.2.0

Input file format:
  see https://github.com/smarco/WFA-paper?tab=readme-ov-file#41-introduction-to-benchmarking-wfa-simple-tests
  Example:
  >ATTGGAAAATAGGATTGGGGTTTGTTTATATTTGGGTTGAGGGATGTCCCACCTTCGTCGTCCTTACGTTTCCGGAAGGGAGTGGTTAGCTCGAAGCCCA
  <GATTGGAAAATAGGATGGGGTTTGTTTATATTTGGGTTGAGGGATGTCCCACCTTGTCGTCCTTACGTTTCCGGAAGGGAGTGGTTGCTCGAAGCCCA
  >CCGTAGAGTTAGACACTCGACCGTGGTGAATCCGCGACCACCGCTTTGACGGGCGCTCTACGGTATCCCGCGATTTGTGTACGTGAAGCAGTGATTAAAC
  <CCTAGAGTTAGACACTCGACCGTGGTGAATCCGCGATCTACCGCTTTGACGGGCGCTCTACGGTATCCCGCGATTTGTGTACGTGAAGCGAGTGATTAAAC

Usage:
  1. Align two sequences from the positional arguments.

        wfa-go [options] <query seq> <target seq>

  2. Align sequence pairs from the input file (described above).

        wfa-go [options] -i input.txt

Options/Flags:
  -N    do not output alignment (for benchmark)
  -a    do not use adaptive reduction
  -g    do not use global alignment
  -h    print help message
  -i string
        input file.
  -m    mem pprof. go tool pprof -http=:8080 mem.pprof
  -p    cpu pprof. go tool pprof -http=:8080 cpu.pprof

Benchmark

Generate datasets with WFA (e634175) or WFA2-lib (v2.3.5):

./bin/generate_dataset -n 100000 -l 1000 -e 0.05 -o l1000-e0.05.seq
./bin/generate_dataset -n 100000 -l 1000 -e 0.10 -o l1000-e0.10.seq
./bin/generate_dataset -n 100000 -l 1000 -e 0.20 -o l1000-e0.20.seq

./bin/generate_dataset -n 500 -l 50000 -e 0.05 -o l50000-e0.05.seq
./bin/generate_dataset -n 500 -l 50000 -e 0.10 -o l50000-e0.10.seq
./bin/generate_dataset -n 500 -l 50000 -e 0.20 -o l50000-e0.20.seq

Commands (all tools use wfa-adaptive heuristic 10,50,1):

# memusg: https://github.com/shenwei356/memusg

# WFA
memusg -t -s "./bin/align_benchmark -i l1000-e0.05.seq -a gap-affine-wfa-adaptive"

# WFA2-lib
memusg -t -s "./bin/align_benchmark -i l1000-e0.05.seq -a gap-affine-wfa --wfa-heuristic wfa-adaptive --wfa-heuristic-parameters 10,50,1"

# WFA-go (this package, binary files are availabe in the release page)
# global alignment && do not output results
memusg -t -s "wfa-go -N -i l1000-e0.05.seq"

csvtk csv2md -t benchmark.tsv -a c,c,c,l,r,r

Results:

Seq-len Seq-num Error-rate Package Time Memory
1000 100000 0.05 WFA1 4.523s 3.21 MB
WFA2 3.597s 4.90 MB
WFA-go 15.424s 9.97 MB
1000 100000 0.1 WFA1 8.031s 2.04 MB
WFA2 6.973s 6.55 MB
WFA-go 41.790s 12.06 MB
1000 100000 0.2 WFA1 15.538s 4.04 MB
WFA2 13.450s 9.24 MB
WFA-go 1m:51s 10.22 MB
50000 500 0.05 WFA1 2.180s 56.12 MB
WFA2 1.481s 107.76 MB
WFA-go 6.107s 86.61 MB
50000 500 0.1 WFA1 4.144s 55.3 MB
WFA2 3.296s 190.32 MB
WFA-go 17.908s 174.62 MB
50000 500 0.2 WFA1 7.574s 81.15 MB
WFA2 6.842s 314.08 MB
WFA-go 48.122s 275.80 MB

Run in a laptop PC, with single-thread.

Reference

Support

Please open an issue to report bugs, propose new functions or ask for help.

License

MIT License

Documentation

Index

Constants

View Source 
const MaskLower32 = 4294967295
View Source 
const MaxSeqLen int = 1<<(32-wfaTypeBits) - 1

MaxSeqLen is the allowed longest sequence length.

View Source 
const OpD = uint64('D')
View Source 
const OpH = uint64('H')
View Source 
const OpI = uint64('I')
View Source 
const OpM = uint64('M')
View Source 
const OpX = uint64('X')

Variables

View Source 
var DefaultAdaptiveOption = &AdaptiveReductionOption{
	MinWFLen:    10,
	MaxDistDiff: 50,
	CutoffStep:  1,
}

DefaultAdaptiveOption provides a default option with parameters from the official repo. attributes.heuristic.min_wavefront_length = 10; attributes.heuristic.max_distance_threshold = 50; attributes.heuristic.steps_between_cutoffs = 1;

View Source 
var DefaultOptions = &Options{
	GlobalAlignment: true,
}

DefaultOptions is the default option

View Source 
var DefaultPenalties = &Penalties{
	Mismatch: 4,
	GapOpen:  6,
	GapExt:   2,
}

DefaultPenalties is from the WFA paper.

View Source 
var ErrEmptySeq error = fmt.Errorf("wfa: invalid empty sequence")

ErrEmptySeq means the query or target sequence is empty.

View Source 
var ErrSeqTooLong error = fmt.Errorf("wfa: sequences longer than %d are not supported", MaxSeqLen)

ErrSeqTooLong means the sequence is too long.

View Source 
var OFFSETS_BASE_SIZE = 128

OFFSETS_BASE_SIZE is the base size of the offset slice.

View Source 
var WAVEFRONTS_BASE_SIZE = 2048

WAVEFRONTS_BASE_SIZE is the base size of the wavefront slice.

Functions

func Op added in v0.3.1 

func Op(op uint64) (byte, uint32)

Op extracts operation type and count.

func RecycleAligner 

func RecycleAligner(algn *Aligner)

RecycleAligner recycles an Aligner object.

func RecycleAlignmentResult added in v0.2.0 

func RecycleAlignmentResult(cigar *AlignmentResult)

RecycleAlignmentResult recycles a CIGAR object.

func RecycleAlignmentText added in v0.2.0 

func RecycleAlignmentText(Q, A, T *[]byte)

RecycleAlignmentText recycle alignment text.

func RecycleComponent added in v0.2.0 

func RecycleComponent(cpt *Component)

RecycleComponent recycles a Component.

func RecycleWaveFront added in v0.2.0 

func RecycleWaveFront(wf *WaveFront)

RecycleWaveFront recycles a WaveFront.

Types

type AdaptiveReductionOption 

type AdaptiveReductionOption struct {
	MinWFLen    uint32
	MaxDistDiff uint32
	CutoffStep  uint32 // not used yet.
}

AdaptiveReductionOption contains the parameters for adaptive reduction

type Aligner 

type Aligner struct {
	M, I, D *Component
	// contains filtered or unexported fields
}

Aligner is the object for aligning, which can apply to multiple pairs of query and ref sequences. But it's not occurrence safe, which means you can't call Align() in multiple goroutines. Instead, you can create multiple aligners, one for each goroutine. Aligner objects are from a object pool, in case a large number of alignments are needed. Just remember to recyle it with RecycleAligner().

func New 

func New(p *Penalties, opt *Options) *Aligner

New returns a new Aligner from the object pool. Do not forget to call RecycleAligner() after using it.

func (*Aligner) AdaptiveReduction 

func (algn *Aligner) AdaptiveReduction(ad *AdaptiveReductionOption) error

AdaptiveReduction sets the adaptive reduction parameters

func (*Aligner) Align 

func (algn *Aligner) Align(q, t []byte) (*AlignmentResult, error)

Align performs alignment with two sequences.

func (*Aligner) AlignPointers 

func (algn *Aligner) AlignPointers(q, t *[]byte) (*AlignmentResult, error)

AlignPointers performs alignment with two sequences. The arguments are pointers.

func (*Aligner) Plot 

func (algn *Aligner) Plot(q, t *[]byte, wtr io.Writer, _M *Component, notChangeToMatch bool, maxScore int)

Plot plots one WFA component as a tab-delimited text table.

A table cell contains the alignment type symbol and the score. Symbols: 

⊕    Unknown,
⟼    Gap open (Insertion)
🠦    Gap extension (Insertion)
↧    Gap open (Deletion)
🠧    Gap extension (Deletion)
⬂    Mismatch
⬊    Match

type AlignmentResult added in v0.2.0 

type AlignmentResult struct {
	// Ops   []*CIGARRecord
	Ops []uint64 // 24-bit null + 8 bit Op + 32-bit N

	Score uint32 // Alignment score

	TBegin, TEnd int // 1-based location of the alignment in target seq, no including flanking clipping/insertion sequences
	QBegin, QEnd int // 1-based location of the alignment in query seq, no including flanking clipping/insertion sequences

	// Stats of the aligned region, no including flanking clipping/insertion sequences
	AlignLen   uint32
	Matches    uint32
	Gaps       uint32
	GapRegions uint32
	// contains filtered or unexported fields
}

AlignmentResult represent a AlignmentResult structure.

func NewAlignmentResult added in v0.2.0 

func NewAlignmentResult(globalAlignment bool) *AlignmentResult

NewAlignmentResult returns a new CIGAR from the object pool.

func (*AlignmentResult) Add added in v0.2.0 

func (cigar *AlignmentResult) Add(op byte)

Add adds a new record in backtrace.

func (*AlignmentResult) AddN added in v0.2.0 

func (cigar *AlignmentResult) AddN(op byte, n uint32)

Add adds a new record in backtrace and set its number as n.

func (*AlignmentResult) AlignmentText added in v0.2.0 

func (cigar *AlignmentResult) AlignmentText(q0, t0 *[]byte, onlyAignedRegion bool) (*[]byte, *[]byte, *[]byte)

AlignmentText returns the formated alignment text for Query, Alignment, and Target. Do not forget to recycle them with RecycleAlignmentText().

func (*AlignmentResult) CIGAR added in v0.2.0 

func (cigar *AlignmentResult) CIGAR(onlyAignedRegion bool) string

CIGAR returns the CIGAR string.

func (*AlignmentResult) Update added in v0.2.0 

func (cigar *AlignmentResult) Update(n uint32)

Update updates the last record.

type Component added in v0.2.0 

type Component struct {
	IsM bool // if it is true, the visualization is slightly different

	WaveFronts []*WaveFront // WaveFronts
}

Component is the wavefront component, it's a list of wavefronts for different scores. To support fast access, we use a list to them, the nil data means there's no such wavefront for a given score.

func NewComponent added in v0.2.0 

func NewComponent() *Component

NewComponent returns a new Component object. If you do not need it, do not remember to use RecycleComponent() to recycle it.

func (*Component) Delete added in v0.2.0 

func (cpt *Component) Delete(s uint32, k int)

Delete delete an offset of a s and k.

func (*Component) Get added in v0.2.0 

func (cpt *Component) Get(s uint32, k int) (uint32, uint32, bool)

Get returns offset, wfaType, existed.

func (*Component) GetAfterDiff added in v0.2.0 

func (cpt *Component) GetAfterDiff(s uint32, diff uint32, k int) (uint32, uint32, bool)

GetAfterDiff returns offset, wfaType, existed for s-diff and k.

func (*Component) GetRaw added in v0.2.0 

func (cpt *Component) GetRaw(s uint32, k int) (uint32, bool)

GetRaw returns "offset<<wfaTypeBits | wfaType", existed.

func (*Component) GetRawAfterDiff added in v0.2.0 

func (cpt *Component) GetRawAfterDiff(s uint32, diff uint32, k int) (uint32, bool)

GetRaw returns "offset<<wfaTypeBits | wfaType", existed for s-diff and k.

func (*Component) HasScore added in v0.2.0 

func (cpt *Component) HasScore(s uint32) bool

HasScore tells if a score exists.

func (*Component) Increase added in v0.2.0 

func (cpt *Component) Increase(s uint32, k int, delta uint32)

Increase increases the offset by delta. Here delta does not contain the backtrace type

func (*Component) KRange added in v0.2.0 

func (cpt *Component) KRange(s, diff uint32) (int, int)

KRange returns the lowest and highest values of k for score s-diff. Since scores are saved in uint32, if diff > s, s-diff would be a large value. So we have to check the two values first.

func (*Component) Print added in v0.2.0 

func (cpt *Component) Print(wtr io.Writer, name string)

Print lists all offsets for all scores and k values.

func (*Component) Reset added in v0.2.0 

func (cpt *Component) Reset()

Reset clears all existing wavefronts for new using.

func (*Component) Set added in v0.2.0 

func (cpt *Component) Set(s uint32, k int, offset uint32, wfaType uint32)

Set sets an offset with a given backtrace type for a score.

func (*Component) SetRaw added in v0.2.0 

func (cpt *Component) SetRaw(s uint32, k int, offset uint32)

Set sets an offset which has already contain a backtrace type for a score. Here, offsetWithType = offset<<wfaTypeBits | wfaType.

type Options 

type Options struct {
	GlobalAlignment bool
}

Options represents a list of options. Currently it only support global or semi-global alignment.

type Penalties 

type Penalties struct {
	Mismatch uint32
	GapOpen  uint32
	GapExt   uint32
}

Penalties contains the gap-affine penalties, Match is 0.

type WaveFront added in v0.2.0 

type WaveFront struct {
	Lo, Hi  int      // Lowest and Highest k.
	Offsets []uint32 // offset data. We preset 2048 values to avoid frequent append operation.
}

WaveFront is a list of offsets for different k values. We also use the low 3 bits to store the backtrace type.

Since k might be negative and usually the values are symmetrical, we store them like this (ZigZag encoding): 

index: 0,  1,  2,  3,  4,  5,  6
k:     0, -1,  1, -2,  2, -3,  3

if the value is 0, it means there's no records for that k.

func NewWaveFront added in v0.2.0 

func NewWaveFront() *WaveFront

NewWaveFront creates a new WaveFront object. If you do not need it, do not remember to use RecycleWaveFront() to recycle it.

func (*WaveFront) Delete added in v0.2.0 

func (wf *WaveFront) Delete(k int)

Delete deletes an offset of a k.

func (*WaveFront) Get added in v0.2.0 

func (wf *WaveFront) Get(k int) (uint32, uint32, bool)

Get returns offset, wfaType, existed.

func (*WaveFront) GetRaw added in v0.2.0 

func (wf *WaveFront) GetRaw(k int) (uint32, bool)

GetRaw returns "offset<<wfaTypeBits | wfaType", existed.

func (*WaveFront) Increase added in v0.2.0 

func (wf *WaveFront) Increase(k int, delta uint32)

Increase increases the offset by delta. Here delta does not contain the backtrace type.

func (*WaveFront) Set added in v0.2.0 

func (wf *WaveFront) Set(k int, offset uint32, wfaType uint32)

Set sets an offset with a given backtrace type.

func (*WaveFront) SetRaw added in v0.2.0 

func (wf *WaveFront) SetRaw(k int, offsetWithType uint32)

Set sets an offset which has already contain a backtrace type. Here, offsetWithType = offset<<wfaTypeBits | wfaType.

func (*WaveFront) String added in v0.2.0 

func (wf *WaveFront) String() string

String lists all the offsets.

Source Files

View all Source files

Directories

Path Synopsis

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.