T2 IRC Log: 2005-05-03

This is the log as captured by an IRC bot in the channel. The statements are those of the individual people and might not neccessarily reflect the policy and legal rules as set forth by the T2 SDE Project.

« prev | next »

--- Log opened Tue May 03 00:00:45 2005
00:08 < jeru> good night all
00:08 -!- jeru [~jeru@p54BFCB6B.dip.t-dialin.net] has left #t2 []
00:13 < valentin> gn8 jeru
01:08 -!- Sparc-kly|G4 [~sparc@66-50-123-218.prtc.net] has joined #t2
01:16 < sparc-kly> hi all
01:29 < sparc-kly> http://www.t2-project.org/
01:29 < sparc-kly> oooops
01:30 -!- ojh [~omer@71-34-253-176.eugn.qwest.net] has joined #t2
02:24 -!- Sparc-kly|G4 [~sparc@66-50-123-218.prtc.net] has left #t2 ["Leaving"]
02:30 -!- sparc-kly [~mubex@66-50-123-218.prtc.net] has quit ["Lost terminal"]
02:45 < mnemoc> rehi
02:53 < ojh> Hi. Anyone know the exact trick to getting NAT forwarding working in a ROCKnet config file?
02:53 < mnemoc> i added a 'conduit' rule, but i'm not sure where
02:55 < mnemoc> give me a minute
02:55 < ojh> I am attempting to follow the example in ROCK-handbook that just uses 'forward', but I don't seem to be getting any forwarding.
02:55 < mnemoc> no, forward is not NAT, is just to enable forward between interfaces
02:56 < ojh> IC, NAT would be part of the IPtables?
02:56 < mnemoc> that forward should be enabled on /etc/sysctl.conf, not on rocknet
02:57 < mnemoc> uhm, i haven't commited conduit
02:57 < mnemoc> what do you know, forward a port to a service inside or masquerade?
02:57 < ojh> Uncomment '# net.ipv4.ip_forward = 1' ?
02:57 < mnemoc> s/kknow/want/
02:59 < mnemoc> that sysctl will let you pass packets from one interface to another
02:59 < mnemoc> that is what 'forward' option on rtocknet does
03:00 < ojh> I want basic NAT between my DSL on eth0 and my LAN on eth1.
03:00 < ojh> IC, need to enable the Kernel to conduct forwarding first.
03:01 < mnemoc> yes
03:01 < mnemoc> do that on sysctl.conf
03:01 < mnemoc> and add 'masquerade' option inside the interface you will use to get outside
03:02 < ojh> Does that need any paramiters?
03:03 < mnemoc> none
03:06 < ojh> BRB
03:06 -!- ojh [~omer@71-34-253-176.eugn.qwest.net] has left #t2 []
03:09 -!- ojh [~omer@71-34-253-176.eugn.qwest.net] has joined #t2
03:16 < ojh> mnemoc: Got it. Its working great. :)
03:16 < mnemoc> :)
03:18 < mnemoc> are you using T2 now or just asking here?
03:19 < ojh> mnemoc: Just asking here. I am attempting to install T2 to the laptop but GRUB mysteriously will not boot from the T2 install CD.
03:19 < mnemoc> oh
03:19 < ojh> mnemoc: My main computer is still running ROCK 2.0.0.
03:20 < mnemoc> the last release before the break :)
03:21 < ojh> mnemoc: The very odd thing is that both T2 and Knoppix are using GRUB 0.95 and it works great on Knoppix.
03:21 < ojh> I gotta go, I'll talk about the problem latter.
03:21 < mnemoc> cu
04:05 -!- ojh [~omer@71-34-253-176.eugn.qwest.net] has left #t2 []
05:16 < CIA-9> amery * r8646 /trunk/package/security/vserver/pkg_linux_pre.conf: * fixed vserver extender .conf
06:08 -!- FeR [~FeR@cable200-116-193-23.epm.net.co] has joined #t2
06:10 -!- FeR [~FeR@cable200-116-193-23.epm.net.co] has quit [Client Quit]
07:22 < rxr> re
07:25 -!- mipe [~mipe@k170.ip2.netikka.fi] has joined #t2
07:50 < rxr> mnemoc: aMule 2.0.0
07:51 < rxr> valentin: Asymptote 0.76
07:55 < rxr> GNU Classpath 0.15 released
08:41 < CIA-9> rene * r8647 /trunk/package/graphic/asymptote/ (asymptote.cache asymptote.desc): * updated asymptote (0.75 -> 0.76)
08:42 < CIA-9> rene * r8648 /trunk/package/network/amule/ (amule.cache amule.desc): * updated amule (2.0.0rc8 -> 2.0.0)
09:38 < rxr> koffice yields _a lot_ of warnings ...
09:38 < rxr> at least the latest beta version
10:08 -!- mipe [~mipe@k170.ip2.netikka.fi] has quit ["Leaving"]
10:20 < rxr> koffice -beta looks good
10:28 < CIA-9> rene * r8651 /trunk/target/desktop/config.in: * added portmap to the desktop target
10:49 < rxr> 269 1380 10296 embedded.txt
10:49 < rxr> rapid article drafting ...
11:04 < rxr> 305 1612 12015 embedded.txt
11:04 < rxr> so - this must be enough for now ...
11:06 < rxr> == 10:18:42 =[9]=> Building kde/koffice [1.3.91 2.2.0-alpha].
11:06 < rxr> -> Compiler Cache Hits while pkg build: 100.00% (475 hits, 0 misses)
11:06 < rxr> == 05/03/05 10:57:06 =[9]=> Finished building package koffice.
11:21 < CIA-9> rene * r8652 /trunk/package/kde/ (koffice/koffice.desc koffice-i18n/koffice-i18n.desc):
11:21 < CIA-9> * updated koffice (1.3.5 -> 1.3.91)
11:21 < CIA-9> * updated koffice-i18n (1.3.5 -> 1.3.91)
12:11 -!- _jeru [~jeru@mail.oberlinhaus.de] has joined #t2
12:11 -!- _jeru is now known as jeru
12:11 < jeru> moin
12:12 < mtr_> moin
12:17 < rxr> moin jeru and mtr_
12:57 < jeru> hi rxr
12:57 < rxr> I reported wrong values for my Embedded C++ stuff
12:58 < rxr> I accidently did really used dietlibc, now I get:
12:58 < rxr> -rwxr-xr-x 1 rene users 29516 May 3 12:56 a.out
13:04 -!- jeru is now known as jeru|away
13:16 < valentin> moin
13:17 < rxr> moin
13:17 < mtr_> moin valentin
13:19 < valentin> moin rxr, mtr_
13:22 < rxr> valentin: just voke up ?
13:23 < valentin> nope
13:23 < valentin> but just got up
13:23 < rxr> ah - I already tried to call you two times this morning ... :-)(
13:23 < valentin> hm ?
13:23 < valentin> the phone did not ring
13:25 < rxr> hm - but I'm sure the call succeeded - since after 12 rings or so your f*cking t-com mailbox picked up ...
13:25 < rxr> maybe you could finally disable thta beast, since AFAIK you do not use it anyway, do you ?
13:26 < valentin> well, k listens to it some times
13:26 < rxr> most of the stuff - like embeded article cross-read are done know anyway ...
13:27 < valentin> hello ?
13:27 < valentin> did you just call me ?
13:28 < rxr> nope
13:29 < valentin> did you speak sth on the mailbox ?
13:29 < rxr> nope
13:29 < valentin> hm, i heard your voice
13:29 < valentin> maybe some old junk
13:30 < valentin> i disabled the voicebox of my landline
13:30 < valentin> so you can call now
13:30 < rxr> ah - yes - maybe - it was strange
13:30 < rxr> when I put the phone down it rang again
13:30 < rxr> I picked up and said my name of course ...
13:31 < rxr> after I put the phone down again since nothing was there it rang a again
13:31 < rxr> without anything ...
13:31 < valentin> the computer voice said sth about "warteschleife"
13:31 < valentin> hm
13:31 < rxr> maybe this is on your mailbox
13:31 < valentin> that phone sucks
13:31 < rxr> may I ask again to disable this damn crap, please?
13:31 < valentin> yes - i heard a "hello"
13:31 < valentin> it is disabled now, so just call
13:32 < rxr> e.g. I also called o2 to disable this new notification service where I get a SMS for calls when the cellphone is not in the homezone
13:32 < rxr> since it costs a few cents for the callee ...
13:33 < valentin> what tf ?
13:33 < valentin> these telco's mug you everywhere
13:40 -!- sparc-kly [~sunultra@64.237.132.176] has joined #t2
14:20 -!- mipe [~mika@dsl10040.japo.fi] has joined #t2
14:22 -!- jeru|away is now known as jeru
14:39 -!- Netsplit kornbluth.freenode.net <-> irc.freenode.net quits: valentin, sepp
14:40 -!- Netsplit over, joins: sepp, valentin
14:41 < valentin> hi jeru
14:49 < jeru> hi valentin
15:06 < valentin> damn, it is so warm in my room - that my ibook cooler spins whithout high cpu load
15:14 < rxr> here it is quite fresh today
15:34 < valentin> now i rearranged some doors and windows and it gets cool here, too
15:47 -!- rxr_ [~rene@e178181173.adsl.alicedsl.de] has joined #t2
15:47 -!- Topic for #t2: T2 | 2.1.0-beta4 RELEASED | The next generation of System Development Environments (SDE) | http://www.t2-project.org/
15:47 -!- Topic set by menomc [] [Sat Apr 16 00:33:34 2005]
15:47 [Users #t2]
15:47 [ _Ragnar_] [ jsaw ] [ mtr_] [ rxr_ ] [ valentin]
15:47 [ CIA-9 ] [ mipe ] [ nzg ] [ sepp ]
15:47 [ jeru ] [ mnemoc] [ rxr ] [ sparc-kly]
15:47 -!- Irssi: #t2: Total of 13 nicks [0 ops, 0 halfops, 0 voices, 13 normal]
15:47 -!- [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
15:47 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
15:47 -!- Irssi: Join to #t2 was synced in 11 secs
15:52 -!- jeru [~jeru@mail.oberlinhaus.de] has quit [Remote closed the connection]
15:53 < mipe> wiki TODO is spammed again
15:53 < rxr_> mnemoc: we should get rid of the wiki ASAP
15:53 < mipe> or put it behind login
15:55 -!- rxr [~rene@e178133172.adsl.alicedsl.de] has quit [Read error: 60 (Operation timed out)]
16:25 < valentin> rxr_: please re-nick
16:26 < valentin> oh - i found a bug in the kde graphics engine
16:27 < valentin> if you have an auto hiding task bar, click on an applet in the tray which has a menu. leave the menu open and do nothing untill the task bar disappears ...
16:28 < valentin> (you have to move the mouse outside the menu so the bar disappears)
16:28 < valentin> after that you have remaining graphic junk below the menu
16:29 < valentin> ah - that is the shadow code
16:30 -!- You're now known as rxr
16:46 < mnemoc> hi
16:47 < mnemoc> rxr: i do want to get rid of the wiki
16:47 < valentin> hi mnemoc
16:47 < valentin> mnemoc: ack, but we should wait until the www2 is complete
16:47 < mnemoc> hi valentin, mipe, webmaster^Wjeru
16:47 < valentin> so we can transfer the todo and stuff
16:48 < mnemoc> valentin: i have moved some stuff... but just some
16:50 < valentin> we should create some kind of "knowledge data base" where we put those stuff like xemacs configuration anecdotes etc
16:50 < valentin> rxr: what is PHONY target ? i thought it should be .PHONY ?
16:51 < rxr> think so
16:51 < rxr> maybe typo ?
16:51 < valentin> guess so
16:51 < valentin> mom, i correct
16:52 < mnemoc> .phony set what rules are not files
16:52 < mnemoc> with the dot
16:52 < valentin> mnemoc: i know what .phony target is
16:53 < mnemoc> :)
16:53 < mnemoc> sorry
16:53 < mnemoc> i missread
16:53 < valentin> in rxr's makefile there was a phony without a '.', so i wanted to know if that was intentional or typo :)
16:54 < mnemoc> typo :)
16:56 < mipe> oh how do i get username to www2? didnt see any register stuff
16:56 < mipe> and hi *
16:57 < mnemoc> mipe: jeru haven't added one... give me a minute
16:58 < mnemoc> mipe: full name? mail address?
16:59 < mipe> mnemoc: Mika Pesu, pmika at japo.fi
16:59 < valentin> rxr: why do you have the second all: target in tests where you depend on libutility ?
16:59 < rxr> forgot "dd" I guess ...
16:59 < valentin> ok, i'll "dd" it
16:59 -!- tilix [~tilix@212.116.139.141] has joined #t2
17:00 < valentin> for the religios fanatics: yes, i know "C-k"
17:03 -!- keinek [~Keinek@201.254.14.47] has joined #t2
17:03 < keinek> hi
17:04 < mnemoc> hi keinek
17:06 < valentin> hi
17:10 -!- tilix [~tilix@212.116.139.141] has left #t2 ["Konversation terminated!"]
17:12 -!- _jeru [~jeru@p54BFD613.dip.t-dialin.net] has joined #t2
17:12 -!- _jeru is now known as jeru
17:12 < jeru> re hi :)
17:13 < mnemoc> hi jeru
17:13 < mnemoc> jeru: what is your smtp?
17:13 < jeru> hi mnemoc .... what smtp do you mean ?
17:13 < mnemoc> jeru: to let plone sent mails
17:14 < jeru> mnemoc: it's not configured yet
17:14 < jeru> or have I missed sth ?
17:14 < jeru> mnemoc: do you need it ?
17:15 < mnemoc> jeru: i wanted to configure it ;) .... but don't worry
17:15 < mnemoc> jeru: other stuff has more priority
17:15 < jeru> mnemoc: ack :)
17:16 < mnemoc> jeru: when do you expect to have it usable? i want to drop wiki asap
17:17 < jeru> mnemoc: while looking in my calendar, I guess, it should be available to the end of the week
17:17 < jeru> and doing some minor works on weekend too
17:17 < mnemoc> Error: fontconfig: Didn't find expected font family. Perhaps URW Type 1 fonts need installing? : Times-Roman <-- hate this!
17:18 < jeru> mnemoc: you will get a nice wiki system too
17:19 < mnemoc> $ fc-list Times-Roman
17:19 < mnemoc> Times:style=Bold Italic
17:19 < mnemoc> Times:style=Bold
17:19 < mnemoc> Times:style=Italic
17:19 < mnemoc> Times:style=Regular
17:19 < mnemoc> uhm
17:19 < mnemoc> jeru: thanks :)
17:25 < jeru> mnemoc: btw, what do you use as spam filter in conjunction with qmail ?
17:25 < mnemoc> qmail-scanner
17:25 < valentin> heavy rain here now
17:25 < mnemoc> also here!
17:25 < mipe> freaking cold here!
17:25 < jeru> valentin: oh! not in Zehlendorf so far
17:26 < mipe> but then again i dont one any 30 C weather here
17:26 < jeru> mnemoc: does qmail-scanner works well ?
17:26 < valentin> jeru: it moves southwest : http://www.wetteronline.de/daten/radar/dber/2005/05/031515.gif?68b279216acaba7d0ab53f07230e7c1c&LANG=de
17:26 < jeru> valentin: thx :)
17:26 < mnemoc> jeru: yep, expensive but afortable
17:27 < jeru> mnemoc: so it eats a lot of resources
17:27 < valentin> no: it moves northeast
17:27 < valentin> http://www.wetteronline.de/daten/radar/dber/std.gif?973b8ffada160df5f239eec2825a55b7&LANG=de
17:27 < mnemoc> jeru: not a lot, but starting perl per mail is not cheap
17:28 < mipe> oh what cluster programs comes with t2?
17:28 < mnemoc> none yet
17:28 < mipe> then i have to add some
17:28 < valentin> mipe: which ones are you thinking about ?
17:28 < jeru> mipe/mnemoc: the GFS stuff is on my wishlist
17:29 < mipe> valentin: not sure yet
17:29 < mipe> just asked in case they want t2 cluster at work to run vr lab
17:29 < valentin> just software or kernel level stuff like mosix ?
17:29 < mipe> i havent done any cluster stuff lately,but i'm guessing openmosix?
17:29 < rxr> valentin: svn HEAD looks ok - thanks
17:29 < mnemoc> beowolf? mosix?
17:30 < mnemoc> mosix is easier
17:30 < mnemoc> open*
17:30 < valentin> *mosix sucks
17:30 < jeru> mnemoc: have you used openmosix ?
17:30 < valentin> we have a 24 x 2,7Ghz mosix cluster at work
17:30 < mnemoc> just played with it, no really used it
17:30 < mipe> hmm,have to check these new stuff... its been **** days since my last experiement with openmosix
17:31 < mnemoc> valentin: building T2?
17:31 < valentin> esp some unix IPC and filesystem access stuff does not work very reliable under mosix
17:31 < valentin> mnemoc: FORGET building t2 in mosix
17:31 < valentin> i guess the outcomme will be far from usable
17:32 < valentin> esp in shell scripts with much redirections and piping you get a lot of random io errors
17:32 < mnemoc> oh
17:32 < valentin> single task programms run well and reliable but with a simple prg | sed | grep line you get about one io error in 24 hours
17:32 < mipe> cluster at work would be running mainly math calculations for 3d rendering
17:33 < valentin> we do bioinformatics stuff
17:33 < mipe> unless there is a way to harness 3d gfx cards to cluster
17:33 < mnemoc> jeru: OT: vserver (quota) conflicts squasfs on 2.4
17:34 < rxr> valentin: maybe their 2.6 user-space relocated stuff is beter ...
17:34 < jeru> in case of clustering network services GFS + IPVS should be fine ;)
17:35 < valentin> we have switched our 2.4 openmosix to 2.6 but it still is not very reliable - though it is better than under 2.4
17:35 < valentin> i would prefer kicking mosix and switch to a hand made solution for our calculations
17:35 < valentin> esp. because the load balancer under mosix behaves quite strange
17:35 < valentin> processes migrate far too often between different nodes
17:36 < valentin> and if you have no gigabit ethernet those relocations are quite expensive
17:37 < valentin> the only thing the mosix (idea) is good for, is to create a virtual server where many thin clients can connect to
17:38 < valentin> and because many users want to execute their jobs on our cluster (and most of them are not very skilled in programming / scripting) mosix will stay :(
17:39 < valentin> and: threaded java programms do not migrate at all ! they just stay at the main node
17:39 < mipe> hmm, how does mosix handle,node dropping on fly. stable?
17:39 < valentin> mipe: you can do that. it is more or less stable
17:40 < valentin> but if you plan to build a cluster for a special purpose a hand made solution (maybe with mpi or even shell scripting) is better than mosix
17:41 < valentin> under mosix you have to script around too, if you want to have a constant number of processes running all time
17:41 < jeru> mipe: but perhaps your vr labs application has some builtin job scheduler for distributing the jobs to the nodes
17:42 < valentin> even a simple script can do. i always create scripts that execute job, and when it finishes it gets the next job
17:42 < valentin> then i start 20 instances of that script in a screen
17:42 < valentin> and that is how we do clustering under mosix :(
17:43 < jeru> valentin: are there any other solutions than openmosix around ?
17:43 < valentin> works quite good - besides the fact that the shell scripts sometimes amok due to io errors and stuff (see above)
17:43 < mipe> jeru: it only distributes the resulting images. well mainly because of 5 gfx cards attached to it. but it doesnt currently distribute the calculation jobs as far as i know
17:43 < valentin> jeru: depends on what you want to do
17:44 < mipe> jeru: the idea is to distribute physics and other calculations to cluster
17:44 < valentin> openmosix just bundles n nodes to one virtual multiprocessor system
17:44 < valentin> mipe: if you need realtime i would not use mosix at all
17:45 < valentin> jobs are migrated back and forth between the nodes and sometimes even stop shortly if there is high load
17:45 < mipe> damm have to write up a review of cluster software available and with hand made solutions
17:46 < mipe> well someday
17:46 < valentin> mipe: for special purposes i would highly recommend that
17:46 < valentin> normaly you do not want mosix at all - you just need a communication framework between your nodes
17:47 < jeru> valentin: + a reliable network filesystem
17:51 < rxr> I have a mail regarding my Open Bench from the LLVM people in my inbox
17:51 < rxr> http://llvm.org#
17:53 < rxr> valentin: good news - I think I do not even need to patch dietlibc
17:53 < rxr> just enable a feature !!!
17:53 < rxr> (who the h*ll is maintianer of our dietlibc support ... ?) *g*
17:54 < valentin> rxr: i think it is someone i do not know ,)
17:55 < valentin> some expert like we are would never oversee such things ...
17:55 * valentin hides
17:59 < rxr> definetly ...
18:00 < rxr> hm - is that this "bejier" or however his name is writen
18:00 < rxr> in today's star trek episode?
18:00 * rxr wonders if there is a episode I do not yet know ..
18:01 < jeru> damn trekkies
18:01 * jeru hides too
18:01 < rxr> I'm not a treky ...
18:02 < rxr> jeru: is there s.th. else on TV one can watch ... *g*
18:02 < valentin> btw: seven of nine now plays a role in boston public
18:02 < valentin> not that i watch such things :)
18:02 < jeru> rxr: I don't see tv anymore ;)
18:02 < jeru> watch
18:03 < rxr> I only see seldom stuff in a tiny window on my desktop ,-)
18:03 < rxr> no tv since long ...
18:03 < jeru> rxr: hmmm ... I guess 10 month or so
18:04 < valentin> jreu: wow
18:04 < valentin> jeru even
18:04 < jeru> rxr: just watching some films
18:04 < valentin> what do you do while eating lunch then ? *g*
18:04 < valentin> my egg rolls are finished, bbl
18:04 < rxr> .oO(typing with one hadn)
18:05 < jeru> and I feel better without the "brain sucker"
18:05 < mnemoc> how many fingers?
18:05 < jeru> :)
18:08 < jeru> mnemoc: btw, again thanks for ooo gem ... works well so far :)
18:09 < mnemoc> jeru: great :)
18:10 < rxr> mnemoc: jeru: I think the UI looks a bit messy
18:10 < rxr> the menus old style gray and the icons strangely mixed KDE / GNOME
18:10 < rxr> is this the case on your side, too ?
18:12 < jeru> rxr: seems to be ok ... no "icon mixing"
18:13 < rxr> cool!
18:13 < rxr> work
18:13 < rxr> s
18:13 < jeru> :)
18:13 < rxr> jeru: hm - can you make a screenshost ?
18:13 < rxr> jeru: works as in my C++ dietlibc hackery ...
18:13 < jeru> of course
18:13 < jeru> :)
18:16 -!- keinek [~Keinek@201.254.14.47] has quit [Remote closed the connection]
18:16 < rxr> huh
18:16 < rxr> it is not star trek?
18:16 < rxr> is that deep space nine - /me confused
18:16 < mnemoc> teletoobies?
18:17 < mnemoc> :)
18:21 < jeru> rxr: you've got mail ;)
18:23 < rxr> ,-)
18:30 < valentin> re
18:31 < CIA-9> rene * r8654 /trunk/package/base/dietlibc/dietlibc.conf:
18:31 < CIA-9> * fixed dietlibc to always build with WANT_DYNAMIC for ctor/dtor
18:31 < CIA-9> handling - anything else does only make since for anything but
18:31 < CIA-9> the tiniest example C code ... - also made the SYSCALL selecting
18:31 < CIA-9> slightly mode intuitive ...
19:08 < rxr> valentin: when you run make in the tests after svn up - you will get embeddedSTL binaries really statically linked with dietlibc ,-)
19:08 < rxr> have fun with the tiny beasts
19:08 < valentin> nice
19:09 < valentin> rxr: do i have to rebuild dietlibc ?
19:10 < rxr> ah - yes
19:10 < rxr> when you want to start them you have ,-)
19:10 < valentin> hehe
19:11 < rxr> oh - s.th. is very inefficient ,-)
19:11 < rxr> user 0m0.428s
19:11 < rxr> user 0m0.055s
19:11 < rxr> the first is my diet + embSTL
19:11 < rxr> the later vanilla g++
19:12 < rxr> maybe it is just the very ineffident vector resizing on every element added ,-)
19:12 < valentin> same options ?
19:12 < valentin> you resize on every element ?
19:12 < rxr> for testing
19:12 < rxr> only
19:12 < valentin> std::vector is guranteed to have amortized O(1) allocation
19:12 < rxr> so far my new / delete operators did not yet work - and the allocator segfaulted
19:12 < rxr> so I set the resize to 1 to stress the allocator and new/delete ... ,-)
19:13 < valentin> ok
19:13 < rxr> valentin: of course my implementation will rock - but first I want to get the basics working
19:13 < _Ragnar_> *lol*
19:13 < rxr> a segfaulting every now and then vector has no value
19:14 < rxr> user 0m0.051s
19:14 < rxr> it was the vector
19:14 < rxr> i uncommented
19:14 < rxr> n += 16; // TODO: inteligent block management - for now stress
19:14 < mipe> what are you coding rxt? some commercial stuff?
19:14 < valentin> it should be n *= 2
19:14 < mipe> rxt=rxr
19:15 < valentin> no, he reinvents the wheel ...
19:15 < valentin> :)
19:15 < mipe> :)
19:15 < rxr> mipe: why do you ask about commercial stuff?
19:16 < valentin> rxr is coding a early userspace variant of hotplug
19:16 < rxr> valentin: btw. the GNU std::vector does never shrink - I already implemented shrinking to my std::string
19:16 < rxr> do you think a shrinking std::vector has a value ?
19:16 < mipe> rxr: you where doing some commercial driver stuff while back...
19:16 < valentin> rxr: what does the standard say about resize () ?
19:17 < rxr> I have to reread it ...
19:17 < rxr> it might not be written at resize() - but by some general performance details listed ...
19:17 < rxr> size pop_back or similar does shrink it as well ... ,-)
19:18 < rxr> user 0m0.028s
19:18 < rxr> with:
19:18 < rxr> if (n < size()*2)
19:18 < rxr> n *= 2;
19:19 < valentin> sorry, i do not understand your conditional ?
19:20 < valentin> what is n ?
19:20 < rxr> void reserve (size_type n) {
19:20 < rxr> if (n <= capacity()) return;
19:20 < rxr>
19:20 < rxr> if (n < size()*2)
19:20 < rxr> n = size()*2;#
19:20 < rxr> sorry cut'n pasted defect code ...
19:20 < rxr> user 0m0.031s
19:21 < valentin> ah
19:21 < rxr> valentin: if you want to contribute embSTL code - the basic infrastructure starts to become usable
19:21 < rxr> mipe: I'm implemending a small scale embedded use STL
19:21 < rxr> mipe: mainly for use in conjunctoin with dietlibc ,-)
19:22 < rxr> mipe: mainly for the upcomming T2 bootup code using udev and needing some hardware detection in the earily userspace
19:22 < rxr> and I wanted to demonstrate for a long time that C++ code can be tiny as well and way saner and maintainable
19:22 < valentin> rxr: your code is broken, it should use capacity () instead of size () !
19:23 < valentin> or - mom
19:23 < valentin> hm
19:23 < valentin> ah - alright
19:24 < rxr> well - I changed it to capacity - does only make a tiny difference
19:24 < valentin> the first conditional saves you from useless allocation
19:24 < valentin> rxr: size was correct. first cond. with capazity and second with size. sorry
19:24 < rxr> valgrind does not seem to be functional with dietlibc binaries ...
19:24 < rxr> is it ?
19:25 < valentin> no idea - you use the ppc variant ?
19:25 -!- sepp [~sepp@p213.54.215.189.tisdip.tiscali.de] has quit [Read error: 110 (Connection timed out)]
19:27 < rxr> nope - I'm on my athlon since a few weeks ...
19:27 < rxr> it is more relaxing to use it ,-) faster and better to type on the USB keyboard ...
19:28 < rxr> valentin: if I'm no too miskaten the linux-modmatch with embeddedSTL is fully functional now ...
19:28 < rxr> at lest a diff of the resulting output does not return a difference anymore ,-)
19:30 < valentin> rxr: nice
19:30 < valentin> size ?
19:32 < rxr> 2nd
19:32 < rxr> I correct format errors in scanf printf
19:32 < rxr> then I post the size
19:41 < rxr> 31756 May 3 19:41 linux-modmatch
19:42 < rxr> user 0m0.027s
19:43 < rxr> the dynamic G++ version has this size:
19:43 < rxr> 20520 May 3 19:42 a.out
19:43 < rxr> and this runtime
19:43 < rxr> user 0m0.052s
19:43 < rxr> the static G++ version has this footprint:
19:43 < rxr> 905984 May 3 19:43 a.out
19:44 < rxr> but a better run-time:
19:44 < rxr> user 0m0.040s
19:44 < rxr> ok /me back implementing - rather than timing the baby ,-)
20:16 < valentin> gah - dietlibc rebuilds binutils ???
20:20 < valentin> rxr: do i need the binutils update or is it ok to cancel ?
20:36 < rxr> no need
20:37 < rxr> hm - wow - it is the primary author of LLVM asking about open bench ...
20:37 < valentin> ok, but now your string tests do not work
20:37 < valentin> string1 runs into an infinite loop and string2 segfaults
20:37 < valentin> s/string1/string
20:40 < rxr> string1 and string 2 work here
20:41 < rxr> you are on PPC?
20:41 < valentin> yes
20:41 < rxr> I wake up my iBook and test ...
20:41 < valentin> before i rebuild dietlibc everything crashed
20:42 < rxr> yes of course
20:42 < rxr> because global and static stuff was not constructed
20:42 -!- mnemoc_ [~amery@200.75.27.58] has joined #t2
20:42 < rxr> and thus e.g. the i/o streams where not contructed when << >> operators got executed ...
20:42 < rxr> that resulted in a NULL fstream passed to the C functions ...
20:42 < rxr> with the rebuild dietlibc this should work ifne
20:42 < rxr> fine even
20:42 < rxr> I test on PPC in some seconds
20:42 < rxr> box booting into Linux right now
20:44 -!- mnemoc [~amery@200.75.27.59] has quit [Read error: 104 (Connection reset by peer)]
20:52 -!- sparc-kly [~sunultra@64.237.132.176] has quit ["Leaving"]
21:01 < rxr> valentin: string indeed does inf.-loop on PPC
21:01 < rxr> guess some singned vs. unsigned char madness on PPC or so
21:02 < valentin> :((
21:03 * rxr debugging soon
21:03 < rxr> that valgrind does not work with dietlibc is more :-(((
21:03 < rxr> at least it does not work for me
21:04 < rxr> dietlibc-0.28# grep -ri valgrind *
21:04 < rxr> CHANGES: fix memory leak in regfree (thanks, Valgrind!).
21:04 < rxr> but there must be some way to get it working
21:04 < rxr> ideas welcome
21:06 < rxr> ehrm
21:06 < rxr> holy sh*t
21:06 < rxr> in gdb or with strace string runs normally ...
21:06 < rxr> is this some PPC risk instruction fetch madness ?!?!?
21:07 < rxr> s/risk/RISC/ of course
21:09 -!- sepp [~sepp@p213.54.203.84.tisdip.tiscali.de] has joined #t2
21:09 < sepp> hi
21:12 -!- jeru [~jeru@p54BFD613.dip.t-dialin.net] has quit ["using sirc version 2.211+KSIRC/1.3.12"]
21:14 < sepp> the bootdisk kernel has no jfs, xfs support
21:14 < rxr> this might be
21:15 < rxr> for 2.2 I plan to fix this by making everything modular and thus be able to support anything on an equal leval
21:17 < sepp> CONFIG_XFS_FS, CONFIG_JFS_FS both =m in the default kernel config
21:17 < rxr> trunk or 2.1 ?
21:18 < sepp> 2.1
21:18 < rxr> I did changes on trunk in the last weeks ...
21:18 < rxr> ah
21:18 < rxr> well - it might very well be to not make the kernel too far
21:18 < rxr> see the mails on the list about that topic
21:18 < rxr> consult with mnemoc_ regarding 2.1 - for 2.2 we'll have all FSs in a initramfs
21:18 < sepp> hmm but then it should be disabled in stone
21:20 < rxr> valentin: it is a dietlibc bug - with the ctor / dtor handling I just enabled
21:20 < rxr> the program is restarted infinitly ...
21:20 * rxr fixing ...
21:21 < rxr> sepp: does the bootdisk target include XFS/JFS support?
21:21 < sepp> rxr: it has stone options for xfs/jfs but no kernel support
21:22 < sepp> and it has xfs, jfs file utils
21:22 < rxr> IIRC XFS and JFS should be autodetected by grepping /proc/filesystems
21:23 -!- mipe [~mika@dsl10040.japo.fi] has quit ["Leaving"]
21:33 < valentin> rxr: why is it restarted ?
21:34 < rxr> I have no idea yet
21:34 < rxr> the code looks ok
21:34 < rxr> I currently rebuild dietlibc with debugging sysmbols and not stripped
21:34 < rxr> maybe I see more then ...
21:34 < rxr> but it also happens with a trivial C program ...
21:37 < rxr> nice - the debug version works ok ...
21:37 < rxr> maybe really some delay slot? that there should be a nop after a branch instruction or so ...
21:38 < valentin> ouhm - this should not happen with a valid compiler
21:38 < valentin> or is there assembly code ?
21:40 < rxr> of course there is assembly code in the startup phase ...
21:50 < rxr> "kernel.org ran on a generic "white box PC" using a shared T1, to the present where it runs on multiple quad Opterons each with 24 gigabytes of RAM, 10 terabytes of disk space, and a gigabit link to the internet."
21:51 * _Ragnar_ wants one ;)
22:45 < valentin> cu later
22:46 < rxr> cu
22:54 -!- mtr [~michael@p54AFB227.dip0.t-ipconnect.de] has joined #t2
22:55 -!- sepp [~sepp@p213.54.203.84.tisdip.tiscali.de] has quit ["Lost terminal"]
22:56 -!- sepp [~sepp@p213.54.203.84.tisdip.tiscali.de] has joined #t2
23:03 -!- mtr_ [~michael@p54AFB1DD.dip0.t-ipconnect.de] has quit [Read error: 60 (Operation timed out)]
23:03 -!- sparc-kly [~sunultra@64.237.132.176] has joined #t2
23:27 -!- keinek [~Keinek@201.254.9.27] has joined #t2
23:27 < keinek> hi
23:27 < sparc-kly> hi keinek
23:27 < keinek> hi sparc-kly
23:27 < keinek> go to es
23:27 < keinek> :=
23:27 < keinek> :)
23:28 < sparc-kly> :P
23:35 < sepp> good night all :)
23:38 < sparc-kly> here 5:45pm
23:38 < sparc-kly> :D
23:38 * sparc-kly love Puerto Rico
23:38 < sparc-kly> heheh
23:39 -!- sparcky [~sunultra@64.237.132.176] has joined #t2
23:39 < rxr> I phoned quite long with lars and so ... - so my dietlibc debugging got delayed
23:40 < rxr> but I know found what is causing the loop
23:40 < rxr> just not yet why it happens ...
23:43 -!- veki [~veki@bunuel-cpe-49.nat-pool.bgd.sbb.co.yu] has joined #t2
23:47 -!- sparc-kly [~sunultra@64.237.132.176] has quit [Read error: 145 (Connection timed out)]
23:48 < rxr> valentin: work arounded
23:48 -!- sparcky is now known as sparc-kly
23:53 < rxr> valentin: the embeddedSTL stuff runs on PPC now - I commit after dietlibc rebuild with this quickfix^Whack
23:56 < rxr> usb and pci matching show no regression on PPC - work fine as well
23:56 < rxr> although I'm behind my schedule this embedded C++ / STL and hotplug++ stuff is comming along quite wel ...
23:56 < rxr> +l
--- Log closed Wed May 04 00:00:06 2005