continuum - P2P replicated document storage service

	Commit message (Collapse)	Author	Age	Files	Lines
*	blob window grows	tv	9 days	1	-5/+10
\|
*	delta known to sender	tv	9 days	1	-1/+7
\|
*	delta propagation	tv	9 days	2	-2/+18
\|
*	one delta at a time	tv	9 days	2	-9/+53
\|
*	time blob messages	tv	9 days	4	-4/+46
\|
*	log why a delta	tv	9 days	1	-0/+6
\|
*	offered deltas lapse	tv	9 days	1	-4/+14
\|
*	throughput from delivered message	tv	9 days	2	-2/+59
\|
*	hop delays	tv	9 days	4	-2/+128
\|
*	benchmark delays by path	tv	9 days	2	-0/+15
\|
*	fix: change publisher stops with the runtime	tv	9 days	1	-3/+8
\|
*	benchmark what sync bytes are	tv	9 days	2	-3/+30
\|
*	changes from a peer never go back	tv	9 days	1	-0/+4
\|
*	older delivered messages leave us level	tv	9 days	1	-1/+3
\|
*	benchmark where twice comes from	tv	9 days	2	-4/+11
\|
*	resend only what is needed	tv	9 days	1	-1/+1
\|
*	benchmark delivered and twice	tv	9 days	2	-0/+30
\|
*	any message taken as delivered	tv	9 days	2	-13/+23
\|
*	slow round back off	tv	9 days	2	-5/+86
\|
*	tree messages need no answer	tv	9 days	3	-6/+40
\|
*	hello again when heard from	tv	9 days	2	-2/+19
\| \| \| \| \| \|	While we wait for a peer's HELLO_REPLY, anything else that comes from it makes our HELLO go again at once, at most once a second, instead of when the backoff says.
*	unanswered pull frees the slot	tv	9 days	2	-1/+23
\| \| \| \| \| \|	When the answer to a pull is overdue (the sync reply timeout: 1 s in the simtests, 10 s in production), the pull gives up the slot, and the next waiting neighbour is pulled at once.
*	overdue sync asked after from then	tv	9 days	2	-4/+4
\|
*	pull after grace	tv	9 days	2	-4/+33
\| \| \| \| \| \| \| \|	Before pulling, a node waits one measured round trip to the announcer (1 s at most) and pulls only what is still missing. If the round trip isn't known yet, it pulls at once, as now. On the clean benchmark links the round trip is a few milliseconds, so those scenarios shouldn't change.
*	announcements answered when lacking	tv	9 days	4	-10/+34
\| \| \| \| \| \| \|	A neighbour now answers an announcement only if it lacks some of the announced heads. Heads that merely differ are no longer enough, since they do while writes go on. The check reads the document, so it runs off the message consumer, as the pull check already does.
*	unanswered members not reached	tv	10 days	2	-14/+43
\|
*	links: only news pass on	tv	10 days	1	-6/+12
\|
*	dashboard: show writer trees	tv	10 days	1	-26/+70
\|
*	writers from authors	tv	10 days	3	-27/+46
\|
*	reach set for the session	tv	10 days	2	-16/+10
\|
*	tree changes had already	tv	10 days	1	-4/+14
\|
*	announcement round counters	tv	10 days	2	-8/+23
\|
*	docs after trees	tv	10 days	1	-5/+6
\|
*	announce what trees miss	tv	10 days	2	-35/+115
\|
*	order members by distance	tv	10 days	2	-26/+59
\|
*	replace eager links by trees	tv	10 days	6	-137/+35
\|
*	add writer trees	tv	10 days	5	-27/+266
\|
*	add writer tree benchmark	tv	10 days	1	-0/+148
\|
*	measure 2-hop neighbourhood	tv	10 days	7	-8/+136
\|
*	dashboard: show eager links on the map	tv	10 days	1	-5/+19
\|
*	add eager link counters	tv	10 days	2	-0/+21
\|
*	introduce eager links	tv	10 days	3	-21/+100
\|
*	skip neighbours with our heads	tv	10 days	2	-15/+37
\|
*	announcements answered only if heads differ	tv	10 days	4	-7/+30
\|
*	retry new peers before probing	tv	10 days	1	-1/+1
\|
*	probe list members	tv	10 days	3	-6/+17
\|
*	links: add failover	tv	10 days	1	-14/+129
\|
*	swarm: links by whom we reach	tv	10 days	2	-12/+25
\|
*	always announce new heads	tv	10 days	2	-45/+9
\|
*	links: when asking for a lot relays, do chunk	tv	11 days	1	-17/+72
\|