Background

Module path is inconsistent with go import path. GO111MODULE=on, as doc said, import "github.com/TheTannerRyan/ring", then get this error:

go: finding module for package github.com/TheTannerRyan/ring
go: downloading github.com/TheTannerRyan/ring v1.1.1
go: found github.com/TheTannerRyan/ring in github.com/TheTannerRyan/ring v1.1.1
go: test1 imports
        github.com/TheTannerRyan/ring: github.com/TheTannerRyan/[email protected]: parsing go.mod:
        module declares its path as: github.com/thetannerryan/ring
                but was required as: github.com/TheTannerRyan/ring 

Solution

Fix the module path:

1. Rename the module path to "github.com/TheTannerRyan/ring": https://github.com/TheTannerRyan/ring/blob/master/go.mod#L1

module github.com/TheTannerRyan/ring
go 1.12 

2. Consider reverting the thetannerryan -> TheTannerRyan rename.
3. Change the doc document to use import "github.com/thetannerryan/ring".

Hello! Sorry for concerning, but can't understand why Marshal.Binary returns [1 0 0 0 0 0 0 3 191 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 32 0 0 0 0 0 0 0 0 32 0 0 128 16 0 0 0 0 0 0 0 36 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 16 0 0 0 0 0 0 0 4 0 0 0 64 0 0 0 0 0 0 16 0 64 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 0] this, fro where this numbers "191, 3,36, ...", Bloom Filter isn't bitset which should give "101010101" this type of bitset? Thank you in advance.

We need to be able to merge multiple different Ring's from different hours of data since we process our data in hourly chunks concurrently. The Marshal/Unmarshal methods will be necessary so we can store already-processed hours in a database so we don't need to re-compute them all of the time.

I added tests to test the Merge and Marshal methods but I made them separate tests. They're still ran by TestMain but they don't affect the stats output.

/cc @TheTannerRyan

unit test below fails

func TestRing(t *testing.T) {
        assert := assert.New(t)

        filter, err := ring.Init(1000, 0.03)
        assert.NoError(err)

        b := []byte{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00}
        d := []byte{0x77, 0x50, 0x52, 0x48, 0x57, 0x32, 0x32, 0x77}

        copy(b, d)
        assert.False(filter.Test(b[:3]))
        assert.Equal(b, d)

        copy(b, d)
        filter.Add(b[:3])
        assert.Equal(b, d)

        copy(b, d)
        filter.Add(b[:3])
        assert.Equal(b, d)

}

part of output:

                Error:          Not equal:
                                expected: []byte{0x77, 0x50, 0x52, 0x1, 0x57, 0x32, 0x32, 0x77}
                                actual  : []byte{0x77, 0x50, 0x52, 0x48, 0x57, 0x32, 0x32, 0x77}

                                Diff:
                                --- Expected
                                +++ Actual
                                @@ -1,3 +1,3 @@
                                 ([]uint8) (len=8) {
                                - 00000000  77 50 52 01 57 32 32 77                           |wPR.W22w|
                                + 00000000  77 50 52 48 57 32 32 77                           |wPRHW22w|
                                 }
                Test:           TestRing

It caused by this line:

h3, h4 := murmur128(append(data, single))

I am an engineer at Elixxir, and we are considering using your bloom filter as part of our system. However, we want to use a more customizable bloom filter, in which we can input our own values for the size of the bit array and the number of hash rounds.Thus, we are presenting this pull request to you for your consideration.