Calculation of common stream numbers

[chan] G3N3R@L
Apr 17 20:49 [raw]

It's possible it's unfinished. The protocol doesn't specify what to do if there are no overlapping streams (addrs would still be forwarded, for example), and calling "intersection" on large sets is CPU intensive. I want to replace the stream list with a bloom filter and you can't do this type of test there very easily (or possibly not at all with any meanignful accuracy). As currently PyBitmessage doesn't allow to create objects in streams other than 1, and won't connect to streams other than 1, it's not something I have on a short term todo list. Peter Surda Bitmessage core developer

[chan] G3N3R@L
Apr 17 21:04 [raw]

Please implement streams soon. I want to see bitmessage scalable so I can promote it to a large audience. The bloom filter idea is very good. A twist on the bloom filter is to do a bit sequence filter instead of a set filter, which would eliminate false positives. Also if you solve the scalability issue there are tech blogs and podcasts that will publicize it. We need more than streams. The structure of the streams must do some bandwidth equalization so a user is not uploading and downloading gigabytes of data per day.

[chan] G3N3R@L
Apr 17 21:17 [raw]

It handles the case where a peer is not interested in any streams.

[chan] G3N3R@L
Apr 17 21:31 [raw]

Common streams are no longer calculated per se. Rather incoming addresses and objects are filtered against streams the local node is interested in while outgoing addresses and objects are filtered against streams the remote node is interested in.

[chan] G3N3R@L
Apr 17 21:43 [raw]

Please implement streams soon. I want to see bitmessage scalable so I can promote it to a large audience. The bloom filter idea is very good. A twist on the bloom filter is to do a bit sequence filter instead of a set filter, which would eliminate false positives. Also if you solve the scalability issue there are tech blogs and podcasts that will publicize it. We need more than streams. The structure of the streams must do some bandwidth equalization so a user is not uploading and downloading gigabytes of data per day.

[chan] G3N3R@L
Apr 17 21:43 [raw]

Please implement streams soon. I want to see bitmessage scalable so I can promote it to a large audience. The bloom filter idea is very good. A twist on the bloom filter is to do a bit sequence filter instead of a set filter, which would eliminate false positives. Also if you solve the scalability issue there are tech blogs and podcasts that will publicize it. We need more than streams. The structure of the streams must do some bandwidth equalization so a user is not uploading and downloading gigabytes of data per day.

[chan] G3N3R@L
BM-2cTgGsfiMJ6875SC8ov7JiUTvoZFKj8h2H

Subject Last Count