Serialization and deserialization

[jefflord-pmg]

Very cool project here.

Serialization and deserialization, I don't see any support for this. To me it seems obvious that would be useful.

I have a dataset of 3M records. Each record has 200 fields. I want to create 200+ filters and store/share/reload them.

Am I missing an easy way that this can already be done?

[jefflord-pmg]

I added this support here, https://github.com/jefflord-pmg/BloomFilter.NetCore, maybe it can be pulled in.

[vla]

Nice idea, but serialization and import/export may conflict, and it might be better to implement import/export functionality as an extension

//Serialization interface
public interface IFilterMemorySerializer
{
	FilterMemory Deserialize(byte[] data);

	byte[] Serialize(FilterMemory filterMemory );
}

[vla]

I added this support here, https://github.com/jefflord-pmg/BloomFilter.NetCore, maybe it can be pulled in.

Consider adding custom implementations of imports and exports

[jefflord-pmg]

@vla thanks so much for replying. Glad to hear this project is not dead. I am fascinated by the concept of bloom filters and finally found a use for them but of course serialization and deserialization are a requirement for me.

Can you elaborate on your comments? Are you suggesting that if this were coded a different way you would accept a pull request for this?

Also, if it were done as an interface, seems like "ISerializableFilter" would be more appropriate?

//Serialization interface
public interface ISerializableFilter
{
	IBloomFilter Deserialize(byte[] data);

	byte[] Serialize(IBloomFilter filter);
}

Then use this only here (so far):

public class FilterMemory : Filter, ISerializableFilter
{
    // current code...

    IBloomFilter ISerializableFilter.Deserialize(byte[] data)
    {
        throw new NotImplementedException();
    }

    byte[] ISerializableFilter.Serialize(IBloomFilter filter)
    {
        throw new NotImplementedException();
    }
}

Once again, great work, I look forward to your reply.

[vla]

What I mean by this is giving the user the ability to customize the deserialization.

Shpped in release 2.5.0

See 3d7ce7a

[jefflord-pmg]

What I mean by this is giving the user the ability to customize the deserialization.

Shpped in release 2.5.0

See 3d7ce7a

Amazing, thank you!

[jefflord-pmg]

@vla Dumb question, is this the correct/expected way to use the new feature?

// serialize the filter to a file
using (var fileStream = new FileStream(@"c:\temp\bloomTest.bin", FileMode.Create, FileAccess.Write))
{
    await ((FilterMemory)bloomFilter).SerializeAsync(fileStream);
}

// create a IBloomFilter with an arbitrary expectedElements since it won't matter
// this is needed since DeserializeAsync is not a static and so we need and instance 
var bloomFilter2 = FilterBuilder.Build(1);

// deserialize the filter from the file
await ((FilterMemory)bloomFilter2).DeserializeAsync(File.OpenRead(@"c:\temp\bloomTest.bin"));

[vla]

yes~

But File.OpenRead requires resource release

[jefflord-pmg]

yes~

Great, and nice catch. I was unsure since the instance method deserialization technique is no something I don't see often, and it's not clear that config parameters will be unused.

Once again, fantastic!