⩖ powrelay.xyz

neat good.

eggs good.

coffee good.

meat good.

although its not necessary to use gzip if it turns out to be slow. i just used it because it achieved better compression than my custom compression algo...

hello!

what kind of classes we could use. well anything. - contains link - contains video - contains image - is less than 20 characters - is more than 20 characters - is over 100 characters - contains line break - pubkey contains two zeroes - pubkey has pow of at least 16 bit - is a reply - is kind 1 note - is kind 7 event - contains link to youtube - is a picture of car - is a picture of cat thats just 15 bits local search. of any classification information.

true anarchists transmit their photos in bmp however.

i think bitwise search could be pretty fast. how? tokenize data. each token is given a bit position. if token exists in data bit is 1, otherwise its 0. now find relevant rows in data with given input tokens. for example search could be: token1, token455, token664 create bitmask from the input data and apply to search data rows to find relevant results.

500 million notes takes: 20000/8*500000000/1000/1000/1000 1,25 TB if we can compress each row to 200 bits with method suggested in previous note, we have compressed search data of 500 million notes to: 200/8*500000000/1000/1000/1000 12,5 GB