todo.txt

move modified async tcp send from code perspective

password with spaces not working

onion address support
	would need to replace everywhere IPs are used in the app

vox not cutting out
closing IM/Chat cutting voice? still transmitting

compartmentalization
	each file/folder has access list
		entry can be 
			group (higher/lower
			specific person
		for group - filekey is encoded with group key
		for person - filekey is encoded with persons public key
	
		file header just contains who has access, no encrypted keys
			except if directory or files are encrypted and we dont want those outside group to see..
		file itself appended with access list and keys

Voice Activated (Headphones)
Push to Talk (Speakers)
Remove Mute

test lan transfer speed, how big does sliding window get?
implement history 

prioritize trust - search / transfer over buddy/profile, etc..

windows key capture hook?

copy clipboard crash


storage view not changing when changing projects on nav bar
	nav bar not completely working
clicking during connect
needs unknown section to show trust nodes that are not resolved yet
multiple clients in chat room - when signed off - all sign off
link trust kill switch
really try to get offline messages sent out
set plain/rich text checked in drop down
set text red when new version of component updated in region, so user knows what to check


warn user files unlocked on close

limit proxy throughput, add proxy breakdown to bandwidth window

link
	options to show unloaded trust fragmetns
	option to show just trusted structure, red filtered

op associations - put in diff sturcture buddyinfo data type/file?


drag/drop IM/Chat window
drag out file from share
IM msg sounds - high, mid, low beep - put in im/chat
Right-click clear IM/Chat log


look at store caching

design realignment
add lockdown to IM manage menu
add move/save profile to manage menu?


share
	show lock/unlock icons for shared files, only delete if path is downloadpath
	put same options in web interface
			
other
	person net open/nat/blocked, not finding local untrusted, losing buddy loc
	lookup not connecting on LAN network?
	storage unlock should be threaded like sharing
	also status if killed
	sortable columns that are either auto, or take a delegate comparer as a parameter
	when switching rooms, put focus on chat inputrooo
	drag/drop file into IM, ask user if like to send
	write mupdate to bootstrap.dat even if it doesnt exist
	status panel icons not transparent?'
	moving folders around in fs, discarding changes, causes crash
	prioritize user generated search/transfer over cache replication s/t
	move load event to the first thing the core thread does on load so that interface loads faster
	save IM mode, buddies or structure view on close
	project coming before primary in nar bar drop down
	dropped packet stats, sliding window stats, add to internal
	ReaderWriterLockSlim
	
quick look at live comm
	login search for highest that is online
	create a persistant connection with them
	keep tabs on higher's higher as well
	**heart beats flow down periodically from top to bottom of live comm
	people always know the current live comm structure
	dont need to worry about highers getting flooded in loc component
	signed location notify is the 'heart beat'

having trouble with getting location offline message to user during logout in the sim

design re-alignment

store caching 
	keep track of when last result was sent
		if greater than a minute, cache entry is up for pruning
	
	if greater than x requests in time period, signal in search result to republish
		if search result tagged with republish, republish result at closest node that didnt have data

	used when high load
	how to keep cache entries up to date?
	cache entries only last for like 5 mins
	requester do the caching, or should requested auto-cache if load too high
		auto-caching might not hit the nodes that requesting nodes would be likely to hit
	index file hashes on network for files over 10 meg, otherwise file can be multisourced from default replicators of file
	always get content once found from node closest to SELF, pull content away from source, ensure it doesn't get flooded
		node publishes data (he is focal point), diseminates to adjacent peers (focal point splits in two, etc..)
	How to prevent publishing overload? ex. 100 members in op all publishing global loc data at same 8 nodes closest to OpID
		need publish search type with acks that return current storage volume, publish closest not full
			
	testing
		compare node load in 64 network with on off
		create auto link routine where nodes make random linkup to someone, gen a root
		create auto post / mail / files / chat use for screen shots
		compare bandwidth at levels
		bandwidth for root node with everyone updating its loc every 2 secs

live
	build test algorithms
	acts as continously active real-time conduit between all members in trusted system (might want to pipe system msgs through this)
	live based on rudp links, if no rudp link, live connect to next canidate, even if loc for someone exists
	rapidly alert up/down links when going offilne so that live structure can be patched quickly wihout nodes thinking location file is still active
	downlink determined by closest sub, if equal, highest rank where 2 nodes split tracks
	uplink determined by highest offline where not challenged
	show trusted/untrusted in live
	ensure when random subset of nodes removed all links/chat rooms still connected through live
	regular heart beats from the top? each link signs (itself and online subs) on way down tree
		everyone in org now knows who's online
		
	instead of lowers continually pinging higher nodes
	every minute highers broadcase location notify to subs
		notifies should be signed in this instance
		receive notify from higher, broadcast to lower
	this also acts as the heart beat of the continually active live comm structure
	
		
bandwidth
	client
		burst protection
			no more than 1kb/s of non rudp packets
		

test
	set fw to open and send back pong that reflects that
	op fw changed t
	nextTryIP the best way 2 solve the problem?
	catch thrown expceptions
	
	in sim show web cache requests / publishes, ensure not too often
	web cache pruning - create a / b / c /d /e / f caches
	
	automated tests - 2open, 1open1nat,  2op1open1nat, 2op2open... , run each until connected
	
	
other
	publisher unknown
	re-publish of local sync is off the hook!

	if file deleted on file sys, should auto go back to locked state
		what if user deletes to recycle bin, warn?

	intead of (accept trust) after name, grey name, and when clicked make header below yellow instead of purple
	
	sim AI
		right click sim 
			Service
				Run test -> select service box	
				Cleanup menu -> select service box
					foreach service selected / run asyncronously
			
	
			chat on / chat off
			Linkup
			Mail -> rand send / reply
			Post -> rand post / reply
			Plan -> 
			
		Clean menu -> cleans cached object
		chat - put chat text in active room, qa - received by members?
		trust - trust / change trust, qa - ensure accepted
		mail - send mail, qa - was received
		board - make post / reply to posts, qa - post reached all in region
		storage - create / add random files / folders with scopes
		profile - randomly change, qa ensure network up to date
		location - change loc name, etc.. qa - network up to date
		schedules, goals, storage, trust, mail can be checked to see if remotes have lastest versions
		
	change goals so goals have id and deligation follows org lines
		that way someone changes positions, switches super / sup, goals remain basically the same
		goal deligation is like a request, once accepted the task remains no matter what happens to above?

	on linkup, trigger inherit from top node
	doesnt seem like trust icon/splash changes are propagated to subs quickly should auto post

	sending large block of text crashes, test other places where string input is unchecked size
	
	list box header color is awful

	chat not selecting default input box
	dont show update when in room

	keep list of keys contacted over broadcast - on rudp connect to key send LAN address
		also for rudp over global proxy, send our local IP specified in global network

	one more stab at security in settings, tray, link changes

	test message board - doesnt seem like refresh is thread safe

	
re-alignment
	leave old network once all subs confirmed to be re-aligned
	publish new tagged trust on re-align confirming current network
	if we get an updated trust from parent with new op key, re-align to that
	automatically cascades down in the background
	on re-align warn if any current subs have not be re-aligned to current yet
	only allow re-alignment if user is root 
	on receieve re-align request, confirm parents parent's up to root have been realigned
	public keys, names, files all stay the same
	anything new created is with new file keys that anyone left out doesnt have access to
	

	
3 trust chain
	adding files to storage causes re-download of unrelated file
	change middle teir file, high level integrates with original verion, but not inherited by subs
	working not saved between runs sometimes
	
global proxying
	have sim print warning in status if global proxy nodes exist in routing table, to ensure they're weeded out
	any way to replace published location with a new version?
	replicating - after knowing all global data types
	ensure 2 global proxied NATs udp hole punch rudp

	rainy day bug 
		2 client network, A client blocked, the B client has open global port, blocked op port
		blocked B op being added to A as open causing tunnel to break, links not synced, packet chaos

next
	use packetstream instead of protocl.writetofile	

	do check tags on node click, or local sync search...

	bandwidth limits / auditing
		when bandwidth chart, figure out why bandwidth quadrupiles with 30% NATs
		really need priority search, some of these startup searches should be done during idle time
		use a packet outbound queue, similar to incoming one used in core thread
			outbound packets throttled there over multiple networks
			implement when bandwidth management is done
		
test
	debug places where searches etc.. are happening, when client knows its not connected
	prevent searches from piling up in disconnected state
	
	combine responsive / established - might want to do with proxying to ensure stuff is published over proxy
	change established to CanRepublish? map out process
	
	ensure search for lastest version of service doesnt get hung up on old versions and return prematurely
	a superior 5 levels up, on other side of network, updating plan should reflect on lower on next location update
	links timing out when traversing strucuture
	link.getlocs / getlocsbelow / use live links?
	link/plan research timer
	**thread per core in sim

other
	news message when connected / disconnected from op
	if news area too small for update, scroll across
	test32 nodes 3 lvls down disposing trust of adjacent link node
	assert in deconstructors that events are set back to null
	GetActiveSessions not threaded for gui, try not to thread dictionary
		
	use last seen date in localSync - show day, not time
	alert user of desyncs detected in localSync, maybe as news alert,
		if local versions are newer than those specified in localSync

	publishing / pruing / re-publishing / replicating uplink requests
	posts, outbound mail re-published on startup?
	use common assist for posts/mail/uplinkReqs?
	crawler - bandwidth, location storage load

	nav bar work with loops?

Interface
	General
		Sorting of listviews
		status panel needs to update as network connects network_statusupdate, trigger news as well
		unload tray icon on exit?
		dont allow nav bar person switch on mail/chat components
		multiline text in tree list row exceeds clip bounds
		
	graphics
		storage - restore
		plans   - edit
		goals	- edit
		tray	- restore, exit
	
	public 
		expand a little more (names associated with sessions/transfers/searches/publish)
		browsing using nav bar needs to trigger searches 
		create function to search through externals for type/id/project, when popping external look for existing window?

	Link
		Online - Show as currently online nodes structured into makeshift hierarchy
			chat/voice uses this hierarchy for high/low - done so that realtime communication is not fragmented when people go offline
		
	News
		crazy refreshes when news not visible and button clicked
		show connected / disconnected
		chat msg (if chat not visible or visited in last minute, or alerted in last minute)
			<project> Chat activityk, Betty: hey group....
		new people online 
		
		more refined, if care about user, do detailed diff between new and old file
			show self only if local didn't make the change, if non-local self change show as white
			detect diff between changed schedule / changed goals or both

	input interface
		move to html from rtf? smaller data format, everything plaintext by default, augmented by html where desired
		customizable style sheet for IM that user can edit
		embed pictures/videos, etc.. in email / posts / chat / IMs
		ability to right-click switch from normal to html view

Components
	General
		Refactor Components with delegates or something, refactor while building visual component
			invokeInterface control->gui, gui->interface control functions handle locking of objects (NO locking from gui code)
		break control into interface comm / control comm so that multiple guis (mobile/mono/wpf) can be implemented easily
		Many actions in gui isnt piped down into control to fire events back to the interface to signal updates for multiple interface instances
		component plug-ins / view plugins for private components
		Remove references to LinkMap, use GetLink
		when component handles exit, assert that all events are emptied of references to debug gui's not releasing event handles
		
		Setting up and editing operation can also change the type/strength of the encryption used
		People can choose the strength/type of the encryption on their profile
		only load secret key on demand somehow, once used replace with zero's so not sitting static in memory
		
	Trust
		components
			links   - getlocs .. check cache functions
			chat    - high level chat is others in loop,
		New function to 'find in link map' so that mail,board,chat can provide the user with an option to 'find' a user
		Key termination command if private key compromised - auto-deletes all data associated with id

	
	IM 
		Saved/Loaded my main IM interface
		public interface
		Log Interface - in chat log
		Crash on close IM, fin being called? unhooking from IMcontrol events? 
		sending file / picture - other side sees link click to accept, starts d/l with progress real time update to window
			result showed inline pic, or link to open file, file put in personal storage?
			
	Chat 
		show unfocused label as red on new messages
		only send full status on connect, otherwise send mini status updates for join/leaves
		figure out why invites don't work a lot, ensure that on connect invites are sent
		host going offline and coming back causes room members to not be synchronized
		double click member - IM, right click  - views, single click - send status, who
		room option - show locations, also for IM, make location font smaller
		check size of signed invite field
		name change should change name in chat logs as well for chat/IM/window title
		implement live rooms / network
		set default cursor in input box
		test that our own link changes are reflected in link changes
		if room custom and no members, kill
		untrusted rooms discussed in chat.txt
		
		in future nodes marked as server can have rooms registered there so only one node to query for full room list
		log does chat/im - setup like yahoo chat log
		on join chat, node asks other nodes for previous singed messages before specified time
		
	Mail 
		soon - sync local mail between users
		tagging messages, filter by tags
		double clicking on name loacates them in link tree in panel
		use labels for organization - for example a project label could be tagged on, so we could filter regarding project x
		delete mail not refreshing view
		test/do re-publishing mail
		*re-publish local outbound un-acked mail on network established

	Board 
		 board scope sub-levels down, auto search highers for posts that have source==target, and show if in scope
		 attaching files
		 detect if post success, if not (not connected) ensure re-post when network.established
		 posts with subs not marked with +
		 ***posts/replies not updating quickly or at all :(
			replies not propagating
		 
	Profile
		time out loading - better searching interface use smoke color and better font, more info searching/dling/etc..
		Search / Transfer process painfully slow in real time mode, analyze
		Rename to Sites? Allow public interlinked sub-pages

	Plans
		old plans not being deleted? seems like extra files in dir, try loading solo and updating local plan file
		shutdown in the middle of the causing load corruption?			
		transfer branch path when goal <-> schedule
		***probably want a custom renderer for the progress bar, default xp has it green
		***schedule slider should respond to mouse scrolls
		
	Storage
		drag into secure storage gives user option to replace what is already there, either on disk or locked
		
		review file references system
		drag files folder over highlights target
		secure copy / paste can use same folder commands as drag/drop
			copy/paste will be used as psuedo-fork to create local copy of groups build to work on in a seperate dir
			ensure really fast, because no 'real' files being created, all dupes in file system

		save storage snapshots (past header file versions) - display at root node	
		signal on comparison page if all files are replicated on network (can go offline and files still exist)

Core
    old projects not cleared from components during savelocal, save / ditch / option?
    each component's minute action should have unique interrupt given by core class
     
	Rudp
		route data over session if desination the same	
		nat/block keep list of recent search ids/kadids so duplicate results arent sent out
		use rudp when available to transmit link / loc data

		investigate LAN max transfer speed / CPU usuage and see what limits current setup imposes
		send packets directly from udp to rudp without switching to core thread
		rudpSession receive packet activate core thread		
		rudp socket send/recv buffer should use packet list instead of static buffers
		rudp session send/ecrypt buffers should be dynamically created for larger sizes

	Search 
		give components priority?? link over location, etc..
		make sure search result locations get to file download
		make sure store locations get to file download direct search -> store response works through firewalls
		both have the ability to be proxied

		record source kaddi with value
		on download try source, id isnt in location list, search for it, need that ID key to connect

		by the time key gets checked for download, profile / link headers should have right key if source is the same, if not gotta search

		search/download priority high (view) normal (network) low (research)
		forward searches to proxies
		if blocked send acks tcp to be forwarded
		analyze contact info sent in acks
		ensure correct contact info gets to transfer control, if blocked/natted/open ensure all work
		make sure download isnt started without sources

		search results need to transfer location info over to downloader
		what does search result from firewalled host look like?
		search results seem to only go out udp
		
		location info should include array of latest versions of each component
			so local area can update for everyone on login to the network
			storage, plans, profile - need latest fetched data of local area
			location data needs to be stored when user is offline for this to work

	udp packet throttle especially for patches
		backlog for packets, for patches, etc.. not for rudp
		packet cache for more than 4k sent in second
		cache should go for 60 seconds make, 240k, bleed cache out, first in first out
		Alert if packet cache has any entries at all
		check in sim if node sending more than x packets per sec
		
	ensure small nets / starting nets can be proxied through global net
		tcp proxies for non-local ops should be bandwidth throttled
		connect status should specify proxied through global
		5 connections in same op but all blocked able to form active op network by proxying throgh global
		
	privacy from untrusted
		mail grey if from untrusted source
		no posts allowed from untrusted
		IM settings allow host to not allow messages from untrusted
		refuse replication of network pieces to blacklisted, inherit blacklist from higher
			dont not replicate to untrusted because then each id needs to be trust verified continually
		heartbeat from top down with unique key, wont talk to 'unknown' nodes in dht without it
		if user is ignored, chat invites are ignored as well
		storage quotas per id to prevent flooding
		public key kill switch	
			if encountered, all local files and profile deleted, app closed
	
Simulator
	track memory / allocated objects
	figure out why IO writes in FiresoftDemo so high, getting 2min/sec through, pretty damn good
	record max udp burst sizes, or do packet size distribution histogram

	***checksigned / idtokey throw exceptions, maybe just return false or something?	
			
quick
	test with null global, even have option for secret at this point?
	ensure callers know what to do when tcpcontrol.sendpacket is overloaded
	use localfilekey for all file names so name diff per computer
	finish self search and then start bucket replication so startup doesnt spit out massive patches
	component dir also encoded with local key / trimmed to 8 chars
	cache files missing from component header - do in load header
			dont delete old path in Cache<component> function, until new header saved: move - save headers - delete old files
	run save headers every 10 secs - every sec is too often, and slows simmy