Timestamps are in GMT/BST.
[0:00] <TheSeeker> latest svn 'safe' yet?
[0:01] <Zothar_Work> I dunno; I'm still on r10906
[0:01] <Zothar_Work> (which is after 996, but before some recent datastore changes)
[0:02] * peacenik (n=chatzill@) Quit (Client Quit)
[0:02] <TheSeeker> I'm on 10903
[0:03] * sanity (n=ian@) has joined #freenet
[0:03] * ChanServ sets mode +o sanity
[0:04] <Zothar_Work> close enough; 10904-10906 were me committing changes for FCP stuff
[0:06] <TheSeeker> I guess I'll just have to check in every few days to see if I need to update...
[0:06] <TheSeeker> (going on a road trip that'll take a few weeks)
[0:23] * Iceman_B^Ltop (n=ice@) Quit ("Swan is hotter than Jasmine ^_^ -=SysReset 2.53=-")
[0:38] * NullAcht15 (n=NullAcht@) Quit (Remote closed the connection)
[0:58] * richard223 (i=joe@) has joined #freenet
[1:01] * hjubal (n=hjubal@) Quit ("Leaving")
[1:21] * drxvii (n=drxvii@) has joined #freenet
[1:22] * rguerra (n=rguerra@) has joined #freenet
[1:23] * rguerra (n=rguerra@) Quit (Client Quit)
[1:24] * rguerra (n=rguerra@) has joined #freenet
[1:24] * JasonMog-sad (n=foo@) has joined #freenet
[1:29] * drxvii (n=drxvii@) has left #freenet
[1:42] * phrosty (n=phrosty@) Quit (Read error: 104 (Connection reset by peer))
[1:43] * phrosty (n=phrosty@) has joined #freenet
[2:00] * stpeter (n=inthelob@) Quit ()
[2:03] <toad_> running away is the worst way to resolve a personality conflict. true or false?
[2:04] <Zothar_Work> false
[2:04] <Zothar_Work> uh, wait
[2:04] <Zothar_Work> it's true it's the worst
[2:04] <Zothar_Work> (I used to have a boss that did that for any conflict after a short "fight")
[2:05] <toad_> yeah, I know people who will just walk off at any disagreement
[2:05] <toad_> and then do what they'd planned to anyway
[2:05] <Zothar_Work> i.e. solves nothing
[2:05] <toad_> I'm talking about something more active, and more personal
[2:05] <Zothar_Work> still solves nothing :)
[2:06] <toad_> he says I interrupt him, I say he drones on and on and ought to be interrupted
[2:06] <toad_> should I walk away?
[2:06] <toad_> from something of mutual benefit?
[2:06] <Zothar_Work> if he won't discuss the matter of his lack of brevity, then warn him and if it continues, then stick to the warning I guess, based on the level of detail available
[2:07] <Zothar_Work> (i.e. warn, then walk away)
[2:07] <toad_> well, i'm happy for him to drone on; i'm not happy for him to tell me off for interrupting his droning on
[2:07] <Zothar_Work> is ignoring him an option? :)
[2:08] <toad_> :)
[2:09] <Zothar_Work> (I have a problem of interrupting people when they're talking, but that's usually 'cause I'm trying to jump to the conclusion and skip a few details that too often don't matter; I guess I like my conversations to be "high bandwidth" sometimes)
[2:09] * rguerra (n=rguerra@) Quit ()
[2:10] <Zothar_Work> well, gotta go, my wife's here; T2UL
[2:10] * Zothar_Work (n=chatzill@) Quit ("Chatzilla 0.9.74 [Firefox 1.5.0.8/2006102516]")
[2:15] * sunrise (n=Administ@) has joined #freenet
[2:17] * sunrise (n=Administ@) Quit (Client Quit)
[2:18] * sunrise (n=Administ@) has joined #freenet
[2:32] * richard223 (i=joe@) Quit ()
[2:33] * TheSeeker is now known as TheSeeker|RoadTr
[2:33] * TheSeeker|RoadTr is now known as TheSeeker|Trip
[2:34] <TheSeeker|Trip> bah, forget it, I'll be back eventually :P
[2:34] <toad_> good night folk
[2:34] * toad_ (n=toad@) has left #freenet
[2:34] * TheSeeker|Trip is now known as TheSeeker|Gone
[2:55] * _ph00 (n=z@) Quit ("Leaving")
[3:04] * MikeW (n=em@) Quit ()
[4:00] * Qt-collective (n=Q-collec@) has joined #freenet
[4:11] * Q-collective (n=Q-collec@) Quit (Read error: 110 (Connection timed out))
[4:32] * _ph00 (n=z@) has joined #freenet
[4:38] * sunrise (n=Administ@) Quit (Read error: 113 (No route to host))
[5:05] * fridim (n=fridim@) Quit (""Putain ils sont o?? ?" -- le petit poucet")
[5:07] * Nelio (n=Nelio_To@) has joined #freenet
[5:09] * fridim (n=fridim@) has joined #freenet
[5:09] * CommanderSky (n=Nelio_To@) Quit (Read error: 110 (Connection timed out))
[5:21] * Guest384 (n=Guest384@) has joined #freenet
[5:22] * edt (n=Ed@) Quit (Read error: 110 (Connection timed out))
[5:23] * edt (n=Ed@) has joined #freenet
[5:23] * ChanServ sets mode +o edt
[5:39] <Iceman_B> whats the fastest way to resolve a "fowardOverloaded" problem?
[5:39] <Iceman_B> increasing bw output?
[6:19] * phrosty (n=phrosty@) Quit (Read error: 104 (Connection reset by peer))
[6:37] * Qt-collective is now known as Q-collective
[6:50] <_ph00> after the great success of the french channel #freenet-fr, some of us decided to start a spanish and an italian channel, #freenet-es and freenet-it. SO, if you want to talk about freenet in spanish or italian, those are the channels. There's not many people on them yet but I'm pretty sure they will grow
[6:51] <_ph00> (btw, I have nothing to do with the -fr channel, I can't even write in french, I said 'after the great success of the french channel' only because I thought that it sounded cool)
[7:26] * phrosty (n=phrosty@) has joined #freenet
[7:30] * eSeong|Sleeping (n=eseong@) has joined #freenet
[7:30] * eSeong|Sleeping (n=eseong@) has left #freenet
[8:00] * fridim (n=fridim@) Quit (sterling.freenode.net irc.freenode.net)
[8:00] * OctobersDark (n=October@) Quit (sterling.freenode.net irc.freenode.net)
[8:00] * jblack (n=jblack@) Quit (sterling.freenode.net irc.freenode.net)
[8:00] * maaxel (i=polaris@) Quit (sterling.freenode.net irc.freenode.net)
[8:00] * Harik (i=harik@) Quit (sterling.freenode.net irc.freenode.net)
[8:00] * fridim (n=fridim@) has joined #freenet
[8:00] * OctobersDark (n=October@) has joined #freenet
[8:00] * jblack (n=jblack@) has joined #freenet
[8:00] * maaxel (i=polaris@) has joined #freenet
[8:00] * Harik (i=harik@) has joined #freenet
[8:08] * ShoeyFighter (n=chertel@) has joined #freenet
[8:12] * sbc (n=sbc@) has joined #freenet
[8:36] * whiterabbit (n=Whiterab@) has joined #freenet
[8:38] * ShoeyFighter (n=chertel@) Quit ("Ex-Chat")
[8:39] * sbc (n=sbc@) Quit ("Ex-Chat")
[9:10] * sanity (n=ian@) Quit ()
[9:12] * whiterabbit (n=Whiterab@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * jblack (n=jblack@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * OctobersDark (n=October@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * maaxel (i=polaris@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * Harik (i=harik@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * fridim (n=fridim@) Quit (sterling.freenode.net irc.freenode.net)
[9:12] * whiterabbit (n=Whiterab@) has joined #freenet
[9:12] * fridim (n=fridim@) has joined #freenet
[9:12] * OctobersDark (n=October@) has joined #freenet
[9:12] * jblack (n=jblack@) has joined #freenet
[9:12] * maaxel (i=polaris@) has joined #freenet
[9:12] * Harik (i=harik@) has joined #freenet
[9:13] * MaxxDrv (n=mack@) has joined #freenet
[9:14] * MaxxDrv (n=mack@) Quit (Client Quit)
[9:20] * Nelio (n=Nelio_To@) Quit ("Client exiting")
[9:27] * hopeatikari (n=hopeatik@) has joined #freenet
[9:41] * slinky (n=slinky@) has joined #freenet
[10:47] * Q-collective (n=Q-collec@) Quit (Read error: 104 (Connection reset by peer))
[10:48] * Q-collective (n=Q-collec@) has joined #freenet
[10:56] * _ph00 (n=z@) Quit ("Leaving")
[11:08] * _ph00 (n=z@) has joined #freenet
[11:11] <_ph00> I see lots od swedes here, yet only one is on #freenet-se
[11:11] <_ph00> same for spanish talking ppl, you can use #freenet-es too
[11:11] <_ph00> italians: #freenet-it
[11:12] <_ph00> and frenchmen #freenet-fr, but that was already well known
[11:15] <_ph00> the point with all these non-english channel is making life somewhat easier for those who are not so good at english (I mean, even worse than myself) so, if your friends don't want to start a node because they think they'd need to speak english on irc, and they don't feel like doing that... well, we're trayig to get some non-english channels started
[11:17] <_ph00> btw, we don't have a german channel yet, and I see many finns and poles too: just do it. Start your own language channel and divert on those channel anyone who seems to be more confortable with their native languages
[11:51] * JustMe (i=JustMe_@) Quit ()
[12:22] * Guest384 (n=Guest384@) Quit ("PJIRC forever!")
[12:35] * hopeatikari (n=hopeatik@) Quit ("Advanced IRC 2.0.4 - don't ask, don't tell")
[12:37] * toad_ (n=toad@) has joined #freenet
[12:37] * ChanServ sets mode +o toad_
[12:43] * Rust (n=for@) has joined #freenet
[12:48] * BloodyRustY (n=for@) Quit (Read error: 110 (Connection timed out))
[12:55] * SecretPolice (n=anon@) has joined #freenet
[12:58] <toad_> nextgens: when will the next freenet-ext.jar be ready? the one that fixes FEC on linux/amd64, and the TreeMap bug?
[13:01] <SecretPolice> hi there, I'm using build 994 and freenet keeps telling me its downloading 996 but it never finishes. what do i do ?
[13:16] <toad_> SecretPolice: interesting
[13:16] * toad_ will try to download it himself in a minute
[13:24] * railk (n=railk@) has joined #freenet
[13:27] * railk (n=railk@) Quit (Remote closed the connection)
[13:27] <SecretPolice> it's been showing that for the last couple of days
[13:27] * railk (n=railk@) has joined #freenet
[13:28] <toad_> :|
[13:28] <toad_> will have a look
[13:28] <toad_> can you register on https://bugs.freenetproject.org/ and file a bug please?
[13:28] <toad_> or at least give me your email address?
[13:29] <toad_> (you'll have to register with nickserv if you want to send a privmsg :< )
[13:35] <SecretPolice> is there not just some way i can get it to try again ?
[13:35] <toad_> people are reporting that inserts are improved ...
[13:35] <toad_> very strange
[13:41] <nextgens> toad_> when it's ready
[13:41] <nextgens> isn't it the appropriate answer ? :)
[13:43] <nextgens> probably not before the updater is able to update it anyway ;)
[13:43] * nextgens hasn't got much time to spare lately
[13:43] <nextgens> will have more after tomorrow afternoon
[13:44] <nextgens> hopefully
[13:48] <CIA-12> toad * r10924 /trunk/freenet/src/freenet/store/BerkeleyDBFreenetStore.java: Tweak the timeouts.
[13:48] * Zothar_Work (n=chatzill@) has joined #freenet
[13:49] * SecretPolice (n=anon@) Quit ("Chatzilla 0.9.76 [Firefox 2.0/2006101023]")
[13:52] * hopeatikari (i=hopeatik@) has joined #freenet
[13:54] * greycat (i=rfc1413@) has joined #freenet
[14:03] * railk (n=railk@) Quit (Remote closed the connection)
[14:10] * railk (n=railk@) has joined #freenet
[14:18] <CIA-12> toad * r10925 /trunk/freenet/src/freenet/store/BerkeleyDBFreenetStore.java: More datastore fixes.
[14:28] * mozillaman (n=borg@) has joined #freenet
[14:43] <toad_> i need some major testing of 10925
[14:43] <toad_> 996 did some datastore breakage, and i need to release 997 in the near future
[14:44] * nextgens might have broken things on r10922
[14:47] * Zothar_Work fires up his java compiler
[14:48] * toad_ is trying to revert your change
[14:50] * nextgens can probably help
[14:50] <toad_> i'm trying to use the Merge tool in eclipse
[14:50] <nextgens> svn merge -r10922:10921
[14:50] <nextgens> then commit
[14:51] <toad_> so it's from the new one to the old one
[14:51] <nextgens> yep
[14:54] <CIA-12> toad * r10926 /trunk/freenet/src/freenet/node/fcp/ClientPutBase.java: Revert nextgens' change. We don't know how many blocks we will insert until we have created the top level metadata.
[14:55] <toad_> done
[14:55] <toad_> ok...
[14:55] <toad_> there's the updater bug
[14:56] <toad_> but i think the datastore changes are fairly urgent
[14:56] * nextgens wonders if it's an updater bug
[14:56] <nextgens> or an insertion problem ;)
[14:56] <nextgens> it used to work well
[14:56] <toad_> no, i can fetch the file
[14:56] <toad_> it's an updater bug with not being able to fetch the next-but-1 file if the next one fails somehow
[14:57] * nextgens thinks you should spend some time on it before releasing 997
[14:57] <toad_> why?
[14:57] <toad_> the datastore bugs could be a problem
[14:57] * varro (n=varro@) has joined #freenet
[14:57] <nextgens> indeed :/
[14:57] <toad_> on the other hand the solution has had limited testing...
[14:58] <nextgens> and most have probably already updated to 996
[14:58] <Zothar_Work> would inserting 995 work around the problem or is 995's insert fubar'd or not really help?
[14:58] <nextgens> and done the migration
[14:58] <toad_> 995 is broken
[14:58] <toad_> i could insert 996 as 995 of course...
[14:58] <Zothar_Work> that'd might not be a bad idea
[14:58] * aptanapaul (n=aptanapa@) has joined #Freenet
[15:02] <CIA-12> toad * r10927 /trunk/freenet/src/freenet/support/io/TempBucketFactory.java: paranoia
[15:03] <CIA-12> toad * r10928 /trunk/freenet/src/freenet/node/Version.java: (log message trimmed)
[15:03] <CIA-12> 997:
[15:03] <CIA-12> - lots of datastore fixes
[15:03] <CIA-12> - fix CSS comments (was causing internal error)
[15:03] <CIA-12> - update initial version of indexes (should speed up fetching them on new nodes)
[15:03] <CIA-12> - minor bugfix in SizeUtil
[15:03] <CIA-12> - wording fix on queue page
[15:03] <toad_> nextgens: most active users will have updated manually by now, but the auto-update is broken, so many won't have
[15:03] <toad_> i'll insert 998 as 995, when i've fixed the auto-update :)
[15:04] * aptanapaul (n=aptanapa@) Quit (Client Quit)
[15:06] <Zothar_Work> make up your mind so I know which latest version to test... :)
[15:08] <toad_> nextgens: there's a guy on frost having problems uninstalling
[15:09] <nextgens> on windows I guess ?
[15:09] <nextgens> "known issue"
[15:27] <toad_> Zothar_Work: please test 997 for now
[15:27] <Zothar_Work> it's up and running
[15:27] <toad_> any nasties in the wrapper.log
[15:27] <toad_> ?
[15:27] <Zothar_Work> nope
[15:46] * hopeatikari (i=hopeatik@) Quit ("Advanced IRC 2.0.4 - don't ask, don't tell")
[15:50] * NullAcht15 (n=NullAcht@) has joined #freenet
[15:50] * railk (n=railk@) Quit (Remote closed the connection)
[16:01] * Rust (n=for@) Quit ()
[16:13] * MineHaunter (n=van85@) Quit ("brb")
[16:17] * MineHaunter (n=van85@) has joined #freenet
[16:28] * vulk (n=chatzill@) has joined #freenet
[16:29] * vulk (n=chatzill@) Quit ("Chatzilla 0.9.76 [Firefox 2.0/2006101023]")
[16:32] * mazza[W] (i=mazzanet@) Quit (Remote closed the connection)
[16:36] * sbc (n=sbc@) has joined #freenet
[16:42] * zhaan (n=zhaan@) has joined #freenet
[16:43] * zhaan (n=zhaan@) Quit (Client Quit)
[16:45] * Caco_Patane (n=caco@) has joined #freenet
[16:45] <Caco_Patane> Hallo
[16:55] * sbc (n=sbc@) Quit ("Ex-Chat")
[17:07] <toad_> hallo!
[17:09] * sanity (n=ian@) has joined #freenet
[17:09] * ChanServ sets mode +o sanity
[17:22] * SecretPolice (n=anon@) has joined #freenet
[17:25] * Iceman_B^Ltop (n=ice@) has joined #freenet
[17:33] <CIA-12> toad * r10929 /trunk/freenet/src/freenet/clients/http/HTTPRequest.java: Buffer the data properly in HTTPRequest parser. Thanks UniquePerson for profile.
[17:35] <CIA-12> toad * r10930 /trunk/freenet/src/freenet/clients/http/HTTPRequest.java: More buffering
[17:45] * egon2003 (n=fargod@) has joined #freenet
[17:46] <CIA-12> toad * r10931 /trunk/freenet/src/freenet/node/Version.java: Documentation
[17:49] <CIA-12> toad * r10932 /trunk/freenet/src/freenet/node/Node.java: Oops
[17:56] * nextgens thinks that you should document those buffer lengths
[17:57] <SecretPolice> how does the autoupdate actually work ? where does it get the new files from ?
[17:57] <nextgens> or at least if they are arbitrary, say it
[17:57] <nextgens> so that people won't try to figure out what's going on
[17:58] <nextgens> SecretPolice> it fetches updated version of the node from freenet
[17:58] <SecretPolice> the website ?
[17:59] <CIA-12> toad * r10933 /trunk/freenet/src/freenet/node/updater/NodeUpdater.java: Fix updater: Wouldn't fetch a new version if had once fetched the old version.
[18:00] <Zothar_Work> SecretPolice: no, using Freenet's storage and retrieval facilities themselves
[18:00] <Zothar_Work> it updates over itself
[18:01] <Zothar_Work> (update.sh/update.cmd do update from the regular website)
[18:01] <SecretPolice> is update.cmd ok to run on XP if its not updating properly then ?
[18:01] <Zothar_Work> yes
[18:04] * sbc (n=sbc@) has joined #freenet
[18:07] <CIA-12> toad * r10934 /trunk/freenet/src/freenet/node/updater/NodeUpdater.java: Remove extra /'s !
[18:08] * tubbie (n=tubbie@) has joined #freenet
[18:09] * egon2003 (n=fargod@) has left #freenet
[18:16] <SecretPolice> er, i just ran the update.cmd thing and it downgraded my node from build 994 to 993 ???
[18:20] <toad_> auto-update isn't working
[18:20] <toad_> since 994
[18:20] * toad_ is about to fix it
[18:23] * Jflesch_ (n=jflesch@) has joined #freenet
[18:24] * JasonPog (n=foo@) has joined #freenet
[18:26] * jblack (n=jblack@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * OctobersDark (n=October@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * maaxel (i=polaris@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * Harik (i=harik@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * fridim (n=fridim@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * whiterabbit (n=Whiterab@) Quit (sterling.freenode.net irc.freenode.net)
[18:26] * whiterabbit (n=Whiterab@) has joined #freenet
[18:26] * fridim (n=fridim@) has joined #freenet
[18:26] * OctobersDark (n=October@) has joined #freenet
[18:26] * jblack (n=jblack@) has joined #freenet
[18:26] * maaxel (i=polaris@) has joined #freenet
[18:26] * Harik (i=harik@) has joined #freenet
[18:26] * Boefje (i=1234@) has joined #freenet
[18:26] * Jflesch (n=jflesch@) Quit (Read error: 145 (Connection timed out))
[18:27] * JasonMog-sad (n=foo@) Quit (Connection reset by peer)
[18:27] <SecretPolice> my node says IE 7 is a security risk. What is the risk ?
[18:30] <MineHaunter> IE does not honor the Content-Type header sent by the node
[18:30] <MineHaunter> so it renders files as html even if they aren't marked as such and thus not filtered
[18:33] <toad_> which means a malicious freesite can include a webbug and identify your IP address
[18:33] <MineHaunter> a malicios freesite like this: http://127.0.0.1:8888/freenet:CHK@qhg6NvKnWfG~mKm1akieQfc70pqWXvlyqyBsOnyUZr0,F65SIZDVMBjP9U1Rm1qXK~9Yb8nITW9NO3GdJw3orR8,AAEC--8
[18:33] <MineHaunter> :P
[18:36] * SecretPolice_ (n=anon@) has joined #freenet
[18:38] * phrosty (n=phrosty@) Quit ("baseball is wrong: man with four balls cannot walk.")
[18:41] <CIA-12> toad * r10935 /trunk/freenet/src/freenet/ (3 files in 3 dirs):
[18:41] <CIA-12> Updater fixes:
[18:41] <CIA-12> - Show update confirm page even if no formpassword (there isn't one from the update button). Means that the update button will work again.
[18:41] <CIA-12> - Correct the update and revocation URIs: If they have a trailing / (they will), remove it. Means that auto-update will work again.
[18:41] <CIA-12> - Better logging.
[18:42] * Boefke (i=1234@) Quit (Read error: 110 (Connection timed out))
[18:50] <CIA-12> toad * r10936 /trunk/freenet/src/freenet/node/Version.java:
[18:50] <CIA-12> 998:
[18:50] <CIA-12> - Fix the updater
[18:50] <CIA-12> - Buffer data properly in HTTPRequest parser (significantly speeds up uploading files through the browser on the queue page)
[18:50] <CIA-12> - Some comments
[18:57] * phrosty (n=phrosty@) has joined #freenet
[18:59] * SecretPolice (n=anon@) Quit (Connection timed out)
[19:01] * NullAcht15 (n=NullAcht@) Quit (Read error: 104 (Connection reset by peer))
[19:03] * NullAcht15 (n=NullAcht@) has joined #freenet
[19:05] * phrosty (n=phrosty@) Quit ("baseball is wrong: man with four balls cannot walk.")
[19:06] * TheShad0 (i=no@) Quit (Read error: 104 (Connection reset by peer))
[19:16] * Jflesch_ is now known as JFlesch
[19:22] * MikeW (n=em@) has joined #freenet
[19:26] * sanity (n=ian@) Quit ()
[19:35] * timmy2chk (n=Violent@) has joined #freenet
[19:40] <tubbie> toad_: In freenet.io.comm.UdpSocketManager, the _unclaimed list slowly fills up with messages over several hours/days. They only get removed when it reaches MAX_UNMATCHED_FIFO_SIZE which is currently set to 50000. On every call to waitFor the complete list has to be iterated, so this might be a possible reason for the increased cpu usage when running a node for some time.
[19:43] <tubbie> 50000 seems to overdo it a little imho :)
[19:53] * varro (n=varro@) Quit ("leaving")
[19:56] <CIA-12> toad * r10937 /trunk/freenet/src/freenet/io/comm/UdpSocketManager.java: Increase the maximum packet size: PPPoE (1492) minus IP (20) minus UDP (8).
[19:57] <CIA-12> toad * r10938 /trunk/freenet/src/freenet/io/comm/UdpSocketManager.java: Make it GRE minus 28, in fact.
[19:57] <CIA-12> toad * r10939 /trunk/freenet/src/freenet/io/comm/UdpSocketManager.java: Comments
[19:57] <toad_> tubbie: hmmm
[19:58] <toad_> tubbie: you have experimental evidence that it fills up?
[19:58] <toad_> tubbie: it shouldn't...
[19:59] <Zothar_Work> toad_: I've seen it fill on numerous occasions; my commit is responsible for the 50000 number
[19:59] <tubbie> System.out.println'ing the sizes of _unclaimed and _filters. _filters never becomes much higher then 125 or so. _unclaimed steadily increases
[19:59] <toad_> hmmm
[19:59] <toad_> what kind of messages don't get matched?
[19:59] <Zothar_Work> now that I know there's a complete list iteration, perhaps 50000 is an order of 10x or more too high
[20:00] <NullAcht15> toad_: you do know that modem users usually have an MTU of only about half a kb, do you?
[20:00] <toad_> well, it depends on the bandwidth really...
[20:00] <toad_> NullAcht15: in which case even one data packet won't fit
[20:00] <toad_> NullAcht15: not a lot we can do about that
[20:00] <Zothar_Work> dunno, but perhaps path MTU discovery is possible and worth a look?
[20:00] <tubbie> toad: mostly pings... but other stuff too
[20:00] <toad_> Zothar_Work: not from java!
[20:00] <Zothar_Work> :)
[20:00] <NullAcht15> does that mean, essentially, that modem users cannot use freenet?
[20:00] <toad_> tubbie: if you could get a dump of the different packet types and how common they are, that would be interesting
[20:01] <tubbie> kk
[20:01] <toad_> NullAcht15: well... their packets will be sent at ~ 1kB anyway
[20:01] <toad_> and they'll get fragmented
[20:01] <toad_> and this will suck
[20:01] <toad_> because they'll have an increased packet loss
[20:01] <toad_> i dunno what we can do about it though
[20:01] <nextgens> toad_> I'm not sure increasing the size as you did makes sence
[20:02] <toad_> fitting useful data in 576 bytes is hard
[20:02] <nextgens> keep in mind that it will increase the time needed to get it sent
[20:02] <toad_> although useful in the long run if we want to pretend to be realplayer :)
[20:02] <Zothar_Work> toad_: perhaps you know this differently than nextgens when I asked; does maximum packet size determine maximum message size on the wire or can message be fragmented?
[20:02] <nextgens> you'll lose in terms of latency
[20:02] <toad_> nextgens: if there isn't that much data to be sent, then less data will be sent
[20:02] <toad_> nextgens: i increased it from 1400 to 1448, that's not a big difference
[20:03] <toad_> Zothar_Work: it gets fragmented at the IP layer
[20:03] * nextgens starts his google interview ...
[20:03] <nextgens> bbl
[20:03] <Zothar_Work> OK, so that would handle a 1MB message just fine then?
[20:03] <toad_> really for modem users and so on, we'd need to go to 512-byte transfer blocks
[20:03] <toad_> Zothar_Work: no
[20:03] <toad_> Zothar_Work: it doesn't include retransmission
[20:04] <toad_> Zothar_Work: one 1448 byte gets split into 2 576 byte packets and a 296 byte packet
[20:04] <toad_> on a modem line
[20:04] <toad_> (well, plus headers)
[20:04] <Zothar_Work> do we have a mechanism for doing our own fragments or is that a big part of the binary blobs project?
[20:04] <toad_> if any of those 3 is dropped, the whole packet fails
[20:04] <toad_> no, we don't do any sort of fragmentation
[20:04] <NullAcht15> toad_: do you think it's possible to determine the MTU dynamically?
[20:05] <toad_> i believe some wierd connections have 8xx byte MTU
[20:05] <toad_> modems have 576
[20:05] <toad_> NullAcht15: yes, but in java?
[20:05] <Zothar_Work> perhaps we should do fragmentation???
[20:05] <toad_> NullAcht15: we'd probably have to bundle some binary executable to do PMTU discovery
[20:05] <Zothar_Work> or would it be better to break the payload across several Messages?
[20:06] <toad_> well...
[20:06] <toad_> what we could do is send the data blocks in 512-byte blocks instead of 1024-byte blocks
[20:06] <toad_> however it would still be a squeeze
[20:07] <toad_> you have 28 bytes of UDP/IP headers, then you have a MAC (32 bytes), a sequence number (4 bytes)
[20:07] <toad_> that leaves 4 bytes
[20:07] <toad_> i don't think that's feasible
[20:07] <toad_> we could use even smaller blocks, but that would really suck
[20:08] <Zothar_Work> ah, there's a block size involved...
[20:08] * toad_ checks ... http://wiki.freenetproject.org/NewPacketFormat
[20:08] <toad_> well we could have dynamic block size
[20:08] <toad_> this isn't just about modems, it's also useful for some kinds of stego
[20:08] <toad_> but it's probably a long term thing
[20:08] <toad_> anyway, lets check the current spec...
[20:09] <toad_> 20 IP, 8 UDP, (is that included in the 576 though? at least the former?)
[20:09] <toad_> 4 seqno, 32 MAC
[20:09] <toad_> 1 byte number of acks, 1 byte number of messages, 2 bytes message type, 2 bytes message length
[20:10] <Zothar_Work> MTU in the IP sense would include the 20 IP and 8 UDP, yes
[20:10] <Zothar_Work> Ethernet's MTU is 1500, which is the wire packet size limit I believe
[20:11] <toad_> hmmm, apparently there are routers which only do 576 because originally IP only did 576
[20:11] <toad_> ethernet's 1500 doesn't include the ethernet header; it's 1518 with that
[20:11] <Zothar_Work> ah
[20:13] <toad_> well
[20:13] <toad_> i think TCP does PMTU discovery
[20:13] <toad_> does it?
[20:13] <nextgens> it depends
[20:13] <toad_> that's the only urgent reason to do it in freenet
[20:14] <toad_> and it'd be a PITA
[20:14] <nextgens> btw, most of the time it won't work
[20:14] <nextgens> as most firewalls/routers are blocking icmp packets
[20:14] <toad_> that's true
[20:14] <toad_> it's possible to do it reliably even then but it's slow
[20:15] <toad_> well
[20:15] <nextgens> hence MTU has and will ever be a PITA
[20:15] <toad_> anyway
[20:15] <toad_> freenet WILL suck on modems
[20:15] <Zothar_Work> I believe ICMP is used for TCP, which Java can't use, so we'd have to do a binary search for the limit based on response time or some such; and yeah, PMTU doesn't work everywhere because not everything is PMTU friendly either because of expecting pre-PMTU traffice or whatever
[20:15] <toad_> is the bottom line
[20:15] <toad_> and if some day we need to use stego with small packets (e.g. faking realplayer), we can deal with it at that point :)
[20:15] <Zothar_Work> :)
[20:16] * toad_ notes that you could certainly fit a 512-byte data transfer block in a 660-byte realplayer fake packet
[20:16] <toad_> although it depends how much bogus headers you have to put in
[20:18] <nextgens> usually the payload is maximized on such protocols
[20:18] * nextgens is dissapointed
[20:19] <toad_> ah
[20:19] <nextgens> I had google interviews planned for tonight .... but it seems they are unable to reach me
[20:19] <toad_> 876 bytes = VPN
[20:19] <toad_> that was the problem
[20:19] <toad_> well, 1448 will fit into two 876's
[20:19] <nextgens> there is a major phone suckage happening here
[20:19] <nextgens> :<
[20:19] <toad_> so it's not _so_ catastrophic
[20:20] <toad_> nextgens: some realplayer streams use 660, some use 330
[20:20] <nextgens> depending on the expected latency I guess
[20:20] <toad_> The encryption header of the VPN will take up space in an Ethernet data
[20:21] <toad_> packet, leaving less room for actual data. You may find that small
[20:21] <toad_> files will transfer across a VPN, but not larger files, because of
[20:21] <toad_> fragmentation issues. In this case, lowering the MTU value on the PC to
[20:21] <toad_> something around 1350 (your situation may require larger or smaller
[20:21] <toad_> values) can get things working. You should experiment. You may also
[20:21] <toad_> need to set the MTU size down on any target servers inside the LAN,
[20:21] <toad_> though you should not need to adjust anything on a BorderManager VPN
[20:21] <toad_> server.
[20:21] <toad_> ugh
[20:22] <toad_> Take the highest number that got replies (in our case it was 1418) and add 28 to that to get the MTU (ours was 1446).
[20:23] <toad_> The Maximum MTU for PPTP tunneling VPN using Ethernet in Windows 9x/ME is 1462.
[20:23] <toad_> In Windows XP/2k/2k3, it seems to be limited to 1400 by default.
[20:23] <toad_> ok
[20:23] <toad_> that's a good reason to use 1400
[20:24] <toad_> well 1372 in fact
[20:24] * nextgens doesn't think so
[20:25] <nextgens> I do think that using cisco's recommended value is probably a better idea
[20:25] <nextgens> ie:
[20:25] <nextgens> 1146 iirc
[20:25] <CIA-12> toad * r10940 /trunk/freenet/src/freenet/io/comm/UdpSocketManager.java: 1400-28. And explain _WHY_!
[20:25] <toad_> where do you get that from?
[20:25] <nextgens> who is going to use M$'s PPTP ?
[20:25] * stilllame (n=lee@) has joined #freenet
[20:26] <toad_> nextgens: is it documented on the web somewhere?
[20:26] <nextgens> I'm almost sure it is yes
[20:26] <nextgens> do you want me to google it ?
[20:26] * toad_ will
[20:26] <toad_> and we can more or less comfortably fit a data packet in that
[20:26] <nextgens> I'm not 100% sure of the value though
[20:27] * nextgens ought to login onto a switch and watch for the default configuration of the IOS
[20:27] <toad_> 1024 - payload, 28 - UDP/IP, 32 - MAC, 4 - seqno, 6 - basic packet headers (2 size 2 type 1 acks 1 msg's)
[20:27] * stilllame (n=lee@) Quit ("Leaving")
[20:28] <nextgens> hmm, it's set to 1500 :/
[20:28] <nextgens> maybe we should do some stats first before spending time on it : how many of our packets are reaching the MTU anyway ?
[20:29] * nextgens has never seen "big" packets on the link
[20:29] <toad_> <span class="stellent-normal"> The VPN tunnel encapsulates traffic
[20:29] <toad_> passing through it within an encrypted wrapper, which tends to add
[20:29] <toad_> about 200 bytes to the size of the datagram. In this case, the MTU on
[20:29] <toad_> the VPN tunnel is reduced to 1260 bytes to compensate for the VPN
[20:29] <toad_> header data added to the datagram as it traverses the tunnel.</span>
[20:29] <toad_> http://support.citrix.com/article/CTX110491
[20:29] <toad_> nextgens: lots get close; a single data transfer packet is 1090 bytes
[20:29] <toad_> with no acks and no other messages
[20:35] <toad_> http://technet2.microsoft.com/WindowsServer/en/Library/e6397c5f-13f5-415c-a490-b6533b89fa3e1033.mspx?mfr=true
[20:35] <toad_> Microsoft uses 1280 bytes in its DNS implementation
[20:36] <Iceman_B^Ltop> whoa, its raining updates
[20:37] <toad_> otoh this guy recommends 1450: http://www.via.ecp.fr/via/ml/streaming/2006-04/msg00036.html
[20:40] <nextgens> that's silly
[20:40] <toad_> agreed
[20:40] <nextgens> 1448 maybe
[20:40] <nextgens> but not 1450
[20:40] * toad_ will go with 1280
[20:40] <nextgens> well, it might be a bit short
[20:41] <nextgens> ^^-^^
[20:42] <toad_> well it's a matter of trading slightly more efficiency on the average case against much less efficiency on the increasingly common and geekish case of VPNing
[20:42] * richard223 (i=joe@) has joined #freenet
[20:42] <nextgens> what about 1492 ?
[20:42] <toad_> you think it's a poor tradeoff? we should do 1448?
[20:42] <toad_> 1448 = 1476 (GRE) minus 28
[20:42] <nextgens> 1492 even
[20:42] <toad_> 1464 = 1492 (PPPoE) minus 28
[20:43] <nextgens> yes, that one
[20:43] <nextgens> most people will use PPPoE on the path
[20:43] <richard223> I do
[20:43] <toad_> we can't easily make it configurable
[20:43] * nextgens has PPPoA
[20:43] <toad_> well we could, but we'd have to tell the other side the limit when we connect
[20:43] <nextgens> hmm why not ?
[20:44] * nextgens doesn't get why
[20:44] <toad_> since there are some low-level changes coming up anyway, maybe that's okay
[20:44] <nextgens> are we padding at that level ?
[20:44] <toad_> nextgens: because incoming packets might also be fragmented?
[20:44] * hjubal (n=hjubal@) has joined #freenet
[20:44] <toad_> if they are sending packets to us, they need to know our MSS
[20:44] <nextgens> nah the OS does the defragmentation, doesn't it ?
[20:45] <toad_> IP does end-to-end fragmentation, but it sucks
[20:45] <toad_> because there's no retransmission
[20:45] <nextgens> well that's up to us to handle it ...
[20:45] <toad_> which means freenet 0.7 on a modem will really suck, but there's not much we can do about that
[20:45] * SecretPolice_ (n=anon@) Quit ("Chatzilla 0.9.76 [Firefox 2.0/2006101023]")
[20:46] <toad_> nextgens: right, and it causes much higher % packet loss, which causes much less data to be sent
[20:46] <toad_> so it's best to just have a sensible limit in the first place
[20:46] * toad_ proposes either 1448 or 1280
[20:46] * Caco_Patane (n=caco@) Quit ("BitchX: need we say more?")
[20:46] <toad_> or make it configurable and tell the other side what it is on connect
[20:47] * nextgens doesn't think that that we should do anything about uber/geekish settings
[20:47] <toad_> nextgens: geeks use freenet
[20:47] * nextgens uses 8021q here
[20:47] <nextgens> are you gonna make a special case for me ?
[20:47] <toad_> nextgens: we have a far greater geek population than kazaa, and we always will have
[20:47] <nextgens> (it's trunking ... /me spares you a googleing)
[20:48] * nextgens thinks 1448 would be the best choice
[20:48] <toad_> even though winxp with VPN is 1400-28
[20:48] <nextgens> "don't care" :p
[20:48] <toad_> err winxp with PPTP
[20:49] <nextgens> most people aren't using VPNs to connect to the internet
[20:49] <nextgens> but to connect to their lan through the internet
[20:49] <toad_> true, but what about wireless (for example)?
[20:49] <nextgens> on wireless the MTU is 1500
[20:49] <toad_> if you have a wireless internet connection, there's a good chance you'll be connecting through a VPN
[20:50] <toad_> at least if it has a semi-public configuration
[20:50] <nextgens> I'm not sure we should encourage people to run a node behind such an unstable media
[20:50] <toad_> because it's hard to mix 802.1x with wifidog
[20:50] <nextgens> worksforme(TM)
[20:51] <toad_> why not? wifi _can_ be stable, if it's well engineered, just like most other last-mile technologies
[20:51] <nextgens> people using such things won't engineer it well
[20:51] <nextgens> :)
[20:51] <toad_> nextgens: why not?
[20:51] <toad_> just because you own the cables doesn't mean you're competent
[20:51] <toad_> just because you don't doesn't make you incompetent
[20:52] <richard223> I tried to run freenet today. with XP SP2 IPv6 enabled. didn't work :S
[20:52] <nextgens> well, I'm not sure that many people will connect their nodes through hotspots
[20:52] <nextgens> I do think many will have a wireless AP in their house
[20:52] <nextgens> some might use it
[20:52] <toad_> anyway ...
[20:52] <nextgens> but they would own the wires ;)
[20:53] <nextgens> and could probably set up 8021x, WPA and so on
[20:53] <toad_> MS uses 1280 bytes on their DNS _server_; a public server may want to have a more conservative value than most clients
[20:53] <nextgens> there is no point in having an "open access" AP for a home
[20:53] <toad_> yeah, I'm biased because of my work with BW
[20:53] <toad_> we're about to set up a network where you can either get free limited access, or you can buy unlimited access
[20:54] <toad_> if you want crypto it'll probably have to go over some form of VPN
[20:54] * toad_ will go with 1372 :)
[20:55] <nextgens> :'(
[20:55] <toad_> well, it's 1372 or 1448
[20:55] <toad_> there's not such a huge difference between them, is there?
[20:55] <toad_> we were using 1400 until two hours ago
[20:56] <nextgens> 76B/packet
[20:56] <toad_> nextgens: huh?
[20:56] <nextgens> well, I don't se the point of changing it
[20:56] <toad_> well, i've already changed it
[20:56] <nextgens> why wasn't 1400 appropriate in the first place ? :)
[20:56] <toad_> so i may as well change it to something sensible
[20:57] <toad_> well i didn't want it to be a unique value for freenet, that's part of it
[20:57] <toad_> i wanted to find a standard value
[20:57] * Iceman_B^Ltop (n=ice@) Quit ("Swan is hotter than Jasmine ^_^ -=SysReset 2.53=-")
[20:57] <nextgens> the biggest it is: the widest range it will have .... the less likely we could be fingerprinted with it
[20:58] <nextgens> as most packets won't reach the MTU anyway
[20:58] <nextgens> but if noone of them reach a low MTU, then it's a charateristic
[20:58] <toad_> well, if we used 9200 then i think we could be fingerprinted from that :)
[20:58] <nextgens> agreed :)
[20:58] <nextgens> well, we have to stay within a reasonable boundary: ie <1500
[20:59] <toad_> right
[20:59] <toad_> so we have 1492-28 (PPPoE), 1476-28 (GRE - does anyone use it?), or 1400-28 (windows VPN), or 1280 (MS conservative value for a DNS server)
[21:00] <toad_> nextgens: do you object vehemently to 1372?
[21:00] <toad_> nextgens: strongly enough to continue the conversation?
[21:01] <nextgens> not if it prevents you from working on the updater
[21:01] <nextgens> ;)
[21:01] <toad_> i'd expect many UDP p2p's to use 1372, for exactly the reason i've given
[21:01] <toad_> although some of them might try to do MTU discovery
[21:01] <nextgens> but doing it that way "because MS's vpns have a low payload" doesn't please me
[21:02] * AnMaster (n=AnMaster@) has left #freenet
[21:02] <toad_> well, some VPNs have much less than that; 1400 is the maximum MTU, the average might be less
[21:02] <toad_> but it seems a reasonable compromize
[21:02] <toad_> ok, lets do that
[21:02] <toad_> nextgens: should the freenet-ext.jar updater be completely separate from the freenet.jar updater?
[21:03] <nextgens> I think so
[21:03] <nextgens> it should have a lower priority
[21:03] <toad_> so it's just another instance of the class?
[21:03] * richard223 (i=joe@) Quit ()
[21:03] <nextgens> or even better: be triggered when the node asks it to be
[21:03] <nextgens> see the constant in NodeStarter
[21:03] <toad_> nextgens: are there freenet-ext.jar's that will break the whole node if the wrong freenet-stable-latest.jar is installed?
[21:03] * nextgens thinks we should "try to fetch an update" when the node reports a newer version is abiable
[21:04] <nextgens> I dunno
[21:04] <nextgens> everything is possible
[21:05] <toad_> so the principle then is that if we want people to upgrade to a new freenet-ext.jar, we update NodeStarter.RECOMMENDED_EXT_BUILD_NUMBER
[21:05] <toad_> if that is higher than our actual ext build number, then the node starts the ext-updater
[21:05] <nextgens> that's the idea yes
[21:05] <toad_> (on startup)
[21:05] <toad_> ok
[21:06] <toad_> how do we determine the freenet-ext.jar current version?
[21:07] <nextgens> we have a class for it
[21:07] <nextgens> ExtVersion.java or something like that
[21:07] <nextgens> brb
[21:09] <toad_> it's in NodeStarter apparently
[21:17] * hjubal is now known as grillide
[21:25] <toad_> nextgens: the update URI will be the same pubkey, and "ext" instead of "update"
[21:26] <nextgens> ok, fine for me
[21:27] <nextgens> so that revocation would work for both
[21:27] <nextgens> nice :)
[21:27] * whiterabbit (n=Whiterab@) Quit ("KVIrc 3.2.5 Anomalies http://www.kvirc.net/")
[21:36] <toad_> nextgens: well i assumed that would be best...
[21:37] * toad_ is making a wrapper object (NodeUpdaterManager) to solve various problems with the current situation
[21:37] <toad_> but it shouldn't take long to sort this out
[21:43] * grillide (n=hjubal@) Quit (Read error: 110 (Connection timed out))
[21:44] * grillide (n=hjubal@) has joined #freenet
[21:45] * toad_ wonders if there will be some nasty race conditions .. probably ..
[21:56] * greycat (i=rfc1413@) Quit ("This time the bullet cold rocked ya / A yellow ribbon instead of a swastika")
[21:58] * TheShado (i=no@) has joined #freenet
[22:01] * sbc (n=sbc@) Quit ("Ex-Chat")
[22:03] * grillide (n=hjubal@) Quit ("..3 2 1 ???check inignition and may God's love be with you???..")
[22:04] <tubbie> toad_: List of packets in _unclaimed after few hours of running: http://code.bulix.org/zec8u0-23580
[22:04] * Zothar_Work (n=chatzill@) Quit ("Chatzilla 0.9.74 [Firefox 1.5.0.8/2006102516]")
[22:06] <nextgens> is it "aviability" or "availability" ?
[22:06] <nextgens> availability iirc
[22:12] * hjubal (n=hjubal@) has joined #freenet
[22:34] * Bombe (n=droden@) Quit (Remote closed the connection)
[22:34] * nextgens (n=nextgens@) Quit (Remote closed the connection)
[22:34] * nextgens (n=nextgens@) has joined #freenet
[22:34] * mazza[W] (i=mazzanet@) has joined #freenet
[22:35] * ChanServ sets mode +o nextgens
[22:35] * Bombe (n=droden@) has joined #freenet
[22:35] * ChanServ sets mode +o Bombe
[22:42] * s14318572 (n=chatzill@) has joined #freenet
[22:42] * littlefae (n=littlefa@) has joined #freenet
[22:42] <s14318572> hello
[22:42] <littlefae> Hello there
[22:51] * tubbie (n=tubbie@) Quit ()
[22:53] * littlefae (n=littlefa@) has left #freenet
[22:59] * sanity (n=ian@) has joined #freenet
[22:59] * ChanServ sets mode +o sanity
[23:11] * timmy2chk (n=Violent@) Quit (Read error: 54 (Connection reset by peer))
[23:13] * mozillaman (n=borg@) Quit (Remote closed the connection)
[23:25] * timmy2chk (n=Violent@) has joined #freenet
[23:29] * dystopia (n=ph33r@) has joined #freenet
[23:30] * JasonPog (n=foo@) Quit (Read error: 104 (Connection reset by peer))
[23:30] * colione (n=colione@) has joined #freenet
[23:34] * JustMe (i=JustMe_@) has joined #freenet
[23:45] * timmy2chk (n=Violent@) Quit (Read error: 54 (Connection reset by peer))
Irc logs of #freenet : 2008 2007 2006 2005
These logs were automatically created by FreenetLogBot on chat.freenode.net using the Java IRC LogBot.