spanmap

spanmap

A specialized collection type that uses uint64 for the index key and interface{} for the data. It's ideal for entries that have indexes that are clustered together, such as transactions logs, time series, and other monotonically or contiguously ordered data.

You can think of spanmap like a cross between a map and an array. It can have any starting index, dynamically grow, and contain small gaps.

This collection should not be used for data where the indexes have random properties or very large gaps.

Getting Started

Installing

To start using spanmap, install Go and run go get:

$ go get -u github.com/tidwall/spanmap

Example

var m spanmap.Map

m.Set(910003, "3")
m.Set(910001, "1")
m.Set(910004, "4")
m.Set(910002, "2")

for i := m.Min(); i <= m.Max(); i++ {
    println(i, m.Get(i).(string))
}

All operations

func (m *Map) Set(index uint64, value interface{}) interface{}
func (m *Map) Get(index uint64) interface{}
func (m *Map) Delete(index uint64) interface{}
func (m *Map) Len() int
func (m *Map) MinIndex() uint64
func (m *Map) MaxIndex() uint64

Outputs:

910001 1
910002 2
910003 3
910004 4

Performance

Single threaded performance comparing this package to the stdlib map[uint64]interface{}, google/btree, and tidwall/celltree.

$ go test -bench .

-- spanmap --
set/sequential 1,000,000 ops in 65ms, 15,402,400/sec, 64 ns/op
get/sequential 1,000,000 ops in 3ms, 292,561,563/sec, 3 ns/op
set/random     1,000,000 ops in 130ms, 7,672,620/sec, 130 ns/op
get/random     1,000,000 ops in 29ms, 34,351,759/sec, 29 ns/op

-- stdlib map --
set/sequential 1,000,000 ops in 338ms, 2,957,562/sec, 338 ns/op
get/sequential 1,000,000 ops in 84ms, 11,881,738/sec, 84 ns/op
set/random     1,000,000 ops in 299ms, 3,339,927/sec, 299 ns/op
get/random     1,000,000 ops in 84ms, 11,944,527/sec, 83 ns/op

-- btree --
set/sequential 1,000,000 ops in 216ms, 4,623,031/sec, 216 ns/op
get/sequential 1,000,000 ops in 169ms, 5,924,168/sec, 168 ns/op
set/random     1,000,000 ops in 689ms, 1,451,352/sec, 689 ns/op
get/random     1,000,000 ops in 884ms, 1,131,258/sec, 883 ns/op

-- celltree --
set/sequential 1,000,000 ops in 80ms, 12,465,058/sec, 80 ns/op
get/sequential 1,000,000 ops in 104ms, 9,625,261/sec, 103 ns/op
set/random     1,000,000 ops in 254ms, 3,944,502/sec, 253 ns/op
get/random     1,000,000 ops in 318ms, 3,142,246/sec, 318 ns/op

These benchmarks were run on a MacBook Pro 15" 2.9 GHz Intel Core i9 using Go 1.12

Contact

Josh Baker @tidwall

License

spanmap source code is available under the MIT License.