I2P dev meeting, April 3, 2007

Quick recap

  • Present:

Complication, jrandom, tapeworm,

完全な IRC ログ

16:01 < jrandom> 0) hi
16:01 < jrandom> 1) net status
16:01 < jrandom> 2) syndie dev status
16:01 < jrandom> 3) ???
16:01 < jrandom> 0) hi
16:01  * jrandom waves
16:01 < jrandom> weekly status notes are not up yet, but there wasn't much in 'em so we can cover it inline here
16:01 < jrandom> jumping into 1) net status
16:01 < jrandom> things seem to be doing pretty well, no major problems atm.  there were some troubles on the irc servers earlier, but the hardware issues have been resolved (thanks cervantes and postman!)
16:01 < jrandom> there's been some more discussion on zzz's blog regarding the ssu/ntcp ideas - check that out for more info
16:01 < jrandom> i don't have much to add on that... anyone have anything to bring up on 1) net status?
16:04 < jrandom> if not, lets move on to 2) syndie dev status
16:04 < jrandom> some good progress on the desktop gui lately, with a few components propagated back into the tabbed gui as well
16:04 < jrandom> we've still got some work to do, but i use the desktop gui for most everything atm. 
16:04 < jrandom> mk has brought up some more ideas and concerns regarding the desktop gui as well, and as always, read the Syndie dev forum to follow the planning and implementation
16:04 <+Complication> indeed, I can also confirm higher IRC sessions persistence
16:04 < jrandom> w3wt
16:06 <+Complication> Seems like testing it again might be scheduled then (during my last test, I found it a bit... intimidating)
16:07 < jrandom> ah yeah, i added labels to most of the buttons now ;)
16:07 < jrandom> though if you're on windows it still does the vertical button labels wrong (need to write a custom layout for that)
16:07 <+Complication> (especially the lack of labels on the many components)
16:08 < jrandom> but its still not ready for alpha... i can use it because i know what everything does/is suposed to do
16:08 <+Complication> over here it's Linux, but good to know, I guess
16:08 < jrandom> but hopefully in the next week or so
16:09 <+Complication> on the Syndie side, I've been wondering about one issue: could the new syncing code is being overzealous, like attempting too many transfers concurrently?
16:09 <+Complication> s/is being/be
16:09 < jrandom> it'll try 5 concurrent fetches per archive
16:10 < jrandom> (and one async import thread)
16:10 <+Complication> Over here, its failure rate against most archives has seen a dramatic rise from earlier times
16:10 < jrandom> hmm
16:10 <+Complication> It could be that more people are syncing too,  but I'd still hope it possible to hit a spare moment when the archive ain't busy
16:10 <+Complication> "hitting a spare moment" and getting a quality sync done, seems to generally not happen, though
16:10 < jrandom> so various fetches fail saying "connection reset" or other tcp-like error message?
16:11 <+Complication> "socket closed" and whatnot
16:11 < jrandom> ah ok
16:11 <+Complication> I haven't really counted them
16:11 <+Complication> This is of course entirely via I2P
16:11 < jrandom> the servers arent currently that hefty (i think they have very limited handling capacity), and that should get imporved
16:12 < jrandom> also, as you and $nymFormerlyKnownAsAnonymous said, we should retry those kinds of failures
16:12 <+Complication> right, that might help too
16:12 < tapeworm> What are the servers based on?
16:12 < jrandom> but we definitely need that to be rock solid and transparent, of course
16:13 < jrandom> tapeworm: homebrew
16:13 <+Complication> though when I mesured "eepget" performance a while back, comparatively to Syndie, eepget got great performance and reliability
16:13 < jrandom> (about a dozen lines of code)
16:13 <+Complication> it pulled 2 x 9 MB from dev.i2p.net while archive.syndie.i2p kept failing on tiny little messages
16:13 < jrandom> oh, thats not really a fair test though
16:14 <+Complication> different boxes?
16:14 < jrandom> and syndie actually /uses/ eepget to fetch
16:14 < jrandom> fetching from apache is pretty different from fetching lots of small files from a homebrew webserver ;)
16:14 <+Complication> hmm... I should probably log overzealously while syncing then
16:15 <+Complication> indeed, and the difference between servers too
16:17 <+Complication> heh, it seems I managed to initiate a sync in the desktop UI
16:17 <+Complication> a task which proved too hard last time :)
16:17 < jrandom> w3wt :)
16:18 < jrandom> ok, anyone have anything else for 2?  if not, lets jump on over to 3) ???
16:18 <+Complication> I do have the habits of a heavy taskbar user, though, so it will likely take some getting used to
16:18 <+Complication> (I usually have the taskbar on auto-hide)
16:19 < jrandom> well, there's a compile time option to put the desktop gui in a shell rather than fullscreen - we can make that a command line switch instead 
16:19 <+Complication> is the desktop gui, in principle, capable of having a "minimize" button?
16:19 < jrandom> its trouble to make it a runtime change though, as swt doesn't allow gui component reparenting (reliably), and you cant change a shell's trim
16:20 < jrandom> oh, yes, definitely possible - good idea
16:20 <+Complication> which would send it to background without affecting the order in which other windows below it are arranged?
16:20 < jrandom> we can toss that into the control menu (top left) or the task menu (top right)
16:20 <+Complication> Because using alt+tab tends to change that
16:21 <+Complication> (something... like the "show desktop" button I typically like to have on the taskbar near the KDE / Start button)
16:21 <+Complication> (another location may prove better, but something of this effect)
16:22 < jrandom> yeah, we can hide it the same wayt the tabbed gui's minimize works (or we can iconize it like the normal windowing minimize button)
16:22 <+Complication> Even if admittedly, minimize and show desktop are different things - now that I consider more, minimize seems a bit more logical.
16:24 <+Complication> As for syncing errors, I currently have 1 instance of HTTP 504, and 4 instances of "socket closed"
16:24 <+Complication> 2 successes
16:24  * TrevorReznik encounters like 70% socket closed
16:24 < jrandom> zounds
16:24 < jrandom> ok, ill look into that and get an update in there asap
16:27 < jrandom> ok, in 3) ??? - anyone have anything else for the meeting?
16:27 <+Complication> I wish I had, but not yet - webcache app still incomplete, since real life ran me over a little
16:28 < jrandom> damn that reality!
16:28  * Complication will try to get the 15 annoying things sorted out of the way
16:32 < jrandom> wr0d
16:32 < jrandom> ok, if there isn't anything else for the meeting...
16:32  * jrandom winds up
16:33  * jrandom *baf*s the meeting closed