--- Log opened Wed Jul 29 00:00:33 2020
00:40 -!- slivera [~slivera@103.231.88.27] has quit [Remote host closed the connection]
00:43 -!- ThomasV [~thomasv@unaffiliated/thomasv] has joined #utreexo
01:20 -!- ThomasV [~thomasv@unaffiliated/thomasv] has quit [Quit: Leaving]
02:00 -!- belcher [~belcher@unaffiliated/belcher] has quit [Quit: Leaving]
04:17 < dergoegge> i have already started to work on this in #177. my first idea was to append the ttls to the proofs but the ttls are not there yet when writing to proofs.dat. currently i fetch the ttls in serveBlocksWorker from the leveldb and send them to the client. might be slow because of all the leveldb calls, but leveldb has a cache so if two client ask for the same blocks the ttls only get fetched once from disk.
05:41 -!- slivera [~slivera@103.231.88.27] has joined #utreexo
06:15 -!- slivera [~slivera@103.231.88.27] has quit [Remote host closed the connection]
09:03 < adiabat> Hm yeah it would be nice to have them in the proofs instead of db
09:03 < adiabat> maskes reading much easier at the cost of writing being harder
09:04 < adiabat> could have byte offsets saved within leveldb, which wouldn't be too hard; right now there's just one giant proof file
10:57 < dergoegge> adiabat how does the remembering in the pollard work? maybe i dont understand the code but do leaf siblings point to each other via `niece[0]` to "remember" each other?
10:58 < dergoegge> its weird because the new leaf points to itself via niece[0] here: https://github.com/mit-dci/utreexo/blob/1ec540cc5d08692eded88673635dee96aecbd8c7/accumulator/pollard.go#L105
10:58 < dergoegge> but then the nieces get swapped here: https://github.com/mit-dci/utreexo/blob/1ec540cc5d08692eded88673635dee96aecbd8c7/accumulator/pollard.go#L119
11:13 < dergoegge> after looking at  accumulator/todo.txt i think my assumption is correct. maybe niece should be called relative, because they can be children, nieces or siblings?
11:17 < adiabat> yeah niece[0] pointing to itself means "remember"
11:17 < adiabat> I figured it was better than adding a "memorable" bool into the polNode struct because that would push up the size of it by 8 bytes
11:18 < adiabat> and the leaves are the only things you need to remember, and leaves don't have nieces, so the niece[] pointers don't do anything otherwise
11:19 < adiabat> from a root node though yes, niece is weird and more like child
11:19 < adiabat> because roots don't have siblings, so child-of-sibling doesn't make sense...
11:20 < adiabat> though I think throughout the code, we treat a root as it's own sibling, and everything works out that way
11:24 < adiabat> yeah I don't have a better name for niece.  It's basically a niece except for the roots, but the roots are confusing I guess
11:24 < adiabat> I think it's better this way than having root's be a different struct that's not a polNode which has child[2] instead of neice[2]
11:25 < adiabat> that would be possible though.  And maybe not that bad since you're mostly explicitly dealing with roots in different ways than nodes
11:42 < dergoegge> yeah it's better than a bool, but with the current way you also remember the sibling. i think the main problem for me is that the whole datastructure is a little unintuitive. 
11:42 < dergoegge> the main reason for remembering is to reduce the proof size right? there is nothing to be gained by just remembering some leaves but leaving the proofs as big as they currently are?
11:43 < dergoegge> because i think #177 does what you suggested in the meeting but it doesn't speed anything up...
15:07 -!- slivera [~slivera@103.231.88.27] has joined #utreexo
16:11 -!- slivera [~slivera@103.231.88.27] has quit [Ping timeout: 260 seconds]
19:11 -!- slivera [~slivera@103.231.88.10] has joined #utreexo
19:15 < adiabat> I think it needs to be in combination with #176 to see a speedup.
19:16 < adiabat> because it's still hashing all the way up to the root instead of stopping when it gets to a populated node
19:16 < adiabat> also I need to test and merge all these things since they work together! :)
21:28 -!- slivera [~slivera@103.231.88.10] has quit [Remote host closed the connection]
--- Log closed Thu Jul 30 00:00:35 2020