T2 IRC Log: 2004-11-12

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 Fri Nov 12 00:00:01 2004
00:06 -!- praenti [~praenti@mail.obster.org] has joined #t2
00:06 < praenti> hi
00:06 < praenti> rxr or valentin: here?
00:11 < praenti> looks like they sleep already. perhaps i should do the same. cu
00:16 < rxr> praenti: yeah - barely ...
00:16 < rxr> but now I'm in bed ...
00:16 < rxr> feel free to post notes ..
00:16 < rxr> n8 all
00:21 < jsaw> ping rxr
00:22 < jsaw> hi praenti
00:25 < jsaw> mnemoc: awake?
00:41 -!- CIA-8 [~CIA@to.je.spocco.com] has quit [Excess Flood]
00:41 -!- CIA-8 [~CIA@to.je.spocco.com] has joined #t2
01:47 < CIA-8> jsaw * r4671 /trunk/package/base/sysvinit/config.in:
01:47 < CIA-8> * temporary fix for subconfig-init.in of sysvinit
01:47 < CIA-8> (otherwise ROCKCFG_PKG_SYSVINIT_INITSTYLE was not set
01:47 < CIA-8> and thus install_init not found)
01:47 < CIA-8> More: see ML
01:48 < valentin> whoa
01:49 < valentin> i used graphviz to generate the dependency graph according to the cache files
01:49 < valentin> for stage 1 only
01:49 < valentin> and i got a picture over 14000 pixels wide
01:50 < valentin> containing mostly of wild arrows
01:50 < valentin> /containing/consisting
01:50 < _Ragnar_> now run a graph optimizer...;)
01:50 < valentin> well, i would need to create some kind of hasse diagramm
01:50 < valentin> do reduce the number of arrows
01:51 < jsaw> hi valentin
01:51 < jsaw> hi _Ragnar_
01:51 < valentin> but the problem is we have many circular dependencies
01:51 < valentin> hi jsaw, _Ragnar_
01:51 < valentin> the circles where the reason why i wanted to plot that graph
01:53 < valentin> think i will create a hasse diagram by breaking all circles first and then add relevant circle edges afterwards
01:54 < _Ragnar_> heya:)
01:55 < valentin> perhaps i should assume all stage 0 tools as leaves and ignore all outgoing arrows
01:55 < valentin> but that will be incorrect, too
01:56 < valentin> btw: konqueror fails to display such large pictures
02:01 < valentin> at least stage 0 looks reasonable :)
02:13 < valentin> hm, ijust tried the new microsoft search engine
02:14 < valentin> searching for linux handbook gets rocklinux two times in the top ten
02:14 < valentin> both before gentoo
02:15 < valentin> searching for linux windows leeds to strange pages - but no microsoft sponsored bashing links so far
02:17 < valentin> searching for my name produces pure bullshit - compared to the good google hits
02:18 < valentin> btw: this site produces much invalid html ...
02:19 < valentin> esp their assistants
02:19 < valentin> guess it will look ok in internet explorer
02:20 < valentin> "valentin ziegler" gets at least one hit for me, but still much bullshit
02:21 < valentin> running w3c validation over this page is fun - seldom seen so much errors on a big companys site
02:22 < valentin> the css for their assistant is so broken that the css validator is unable to parse it *g*
02:22 < valentin> ok, it is only a beta version
02:23 < valentin> enough search.msn.* for today
02:23 < valentin> think i'll stay with google
02:32 -!- rxr_ [~rene@p213.54.209.35.tisdip.tiscali.de] has joined #t2
02:32 -!- Topic for #t2: T2 | the system development environment | http://www.exactcode.de/t2
02:32 -!- Topic set by valentin [] [Sun Nov 7 16:41:27 2004]
02:32 [Users #t2]
02:32 [ _Ragnar_] [ daja77] [ martin_] [ praenti] [ rxr_ ] [ tschmidt__]
02:32 [ CIA-8 ] [ jsaw ] [ mnemoc ] [ rxr ] [ sparc-kly] [ valentin ]
02:32 -!- Irssi: #t2: Total of 12 nicks [0 ops, 0 halfops, 0 voices, 12 normal]
02:32 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
02:32 -!- [freenode-info] why register and identify? your IRC nick is how people know you. http://freenode.net/faq.shtml#nicksetup
02:32 -!- Irssi: Join to #t2 was synced in 11 secs
02:35 < valentin> running transitive reduction over the dep graph looks much better
02:37 -!- rxr [~rene@p213.54.237.161.tisdip.tiscali.de] has quit [Read error: 60 (Operation timed out)]
02:37 < jsaw> valentin: may I have a glance at it?
02:38 < valentin> ok, i'll upload it
02:38 < jsaw> I've posted this a couple of times today, but I'm so amazed at the low error count so far that I have to:
02:38 < valentin> this tred tool is very smart
02:38 < jsaw> 1322 builds total, 734 completed fine, 7 with errors.
02:39 < valentin> smarter than my ad hoc implementations
02:39 < valentin> there are only two relevant cycles left after the reduction
02:40 < valentin> http://exactcode.de/valentin/stage1.png
02:41 < valentin> this time it even fits into konqueror
02:41 < valentin> but beware: there were some cache files missing
02:41 < valentin> sed: can't read /usr/src/t2-trunk/package/develop/distcc/distcc.cache: No such file or directory
02:41 < valentin> sed: can't read /usr/src/t2-trunk/package/filesystem/squashfs-tools/squashfs-tools.cache: No such file or directory
02:41 < valentin> sed: can't read /usr/src/t2-trunk/package/sparc/silo/silo.cache: No such file or directory
02:42 < jsaw> ./scripts/Config -oldconfig
02:42 < jsaw> ?
02:42 < valentin> and stage != 1 packages will appear as leave nodes
02:42 < jsaw> valentin: run the image thru potrace or autotrace!
02:42 < valentin> what exactly
02:43 < valentin> does that do ?
02:43 < jsaw> convert bitmap into vector graphics
02:43 < jsaw> nice image btw!
02:43 < valentin> i can give you vector output if you want
02:44 < jsaw> no, just wanted to say, that this would be smaller... *g*
02:45 < valentin> oh, i did not take a look at the size - loads fast here *g*
02:45 < valentin> i'll do a postscript
02:45 < jsaw> :) doesn't matter I have flatrate
02:46 < valentin> or do you prefer svg ?
02:46 < jsaw> oh, you are totally right. it's mainly bw, so png should compress it nicely...
02:46 < jsaw> <- stupid
02:46 < valentin> ok, stage1.ps online
02:47 < jsaw> 52k (png)
02:47 < valentin> -rw-r--r-- 1 valentin users 52571 Nov 12 02:43 stage1.png
02:47 < valentin> -rw-r--r-- 1 valentin users 38519 Nov 12 02:52 stage1.ps
02:48 < valentin> vector formats like ps/svg look better when zooming out :)
02:48 < jsaw> what if you remove 00-dirtree?
02:48 < valentin> i'll try - mom
02:51 < valentin> http://exactcode.de/valentin/stage1b.ps
02:51 < valentin> looks almost the same - bash plays the role of dirtree in that graph
02:52 < mnemoc> rehi
02:52 < jsaw> looks more sane to me, probably we should sort 00-dirtree to be the first thing created ever...
02:52 < jsaw> hi mnemoc
02:53 < mnemoc> jsaw: hi, i'm answering your mail right now... that -z should have been -n but wait for the mail :)
02:53 < valentin> i proposted to rxr in private quert that we should store stage info within the dependency lists
02:53 < valentin> query even
02:53 < valentin> but then creation of cache files will be more compilcated
02:54 < valentin> problem is, we have all the dependencies to the toolchain for almost every package
02:54 < valentin> and the toolchain itself depends on the toolchain again
02:54 < valentin> so there are hundreds of loops
02:54 < jsaw> mnemoc: oh... but is it garanteed that the variable has a value if empty before?
02:55 < jsaw> valentin: but we know the subset which is the toolchain... so we can remove it!?
02:56 < valentin> yes, rxr and i thought about that
02:56 < mnemoc> jsaw: that was a hack to avoid transition errors, but i introduced a new one :(
02:56 < valentin> but how would you handle the toolchain itself
02:56 < valentin> ?
02:56 < mnemoc> hi valentin
02:56 < valentin> hi mnemoc btw
02:56 < jsaw> didn't rxr said, such discussion should happen on the ML *hint* *hint* ;)
02:56 < valentin> yes, as soon as i have a robust idea - for the moment i have none
02:57 < jsaw> s/said/say/, damn, I always make this error...
02:57 < valentin> that is why i look into dep graphs to find out what problems arise
02:57 < jsaw> valentin: make a toolchain stage 3, beginning with stage 4, remove all programs of stage 3
02:57 < valentin> i now crunch the whole graph for 1217 packages
02:58 < jsaw> ohoh
02:58 < valentin> jsaw: i thougt about something like that
02:59 < valentin> i want to know how many of the 32355 edges remain after transitive reduction
02:59 < jsaw> on the other hand, the circular deps do not hurt, if a package is built it is built and can be dropped from dep tree genration.. isn't it?
02:59 < valentin> well - different stages may have different dependencies
02:59 < jsaw> <- getting tipsy - I should shut up slowly
02:59 < valentin> hehe
03:00 < jsaw> different stages different dependencies, yes, but still, the pkgs from prev. stages can be simply dropped...
03:00 < valentin> yes that is true i think
03:01 < valentin> i'll find out
03:01 < jsaw> and if there's really some circular dependency due to an optional package, it could be still detected.
03:05 < jsaw> but somehow I think there's none. Or could it really be, that a package is needed to build another package that then enhances the capabilities of the first package? no, or?
03:05 < valentin> i remember whe had such cases in the old drock
03:06 < valentin> but that can only be resolved by using differnet stages anyway
03:06 < jsaw> somehow I feared it could be true .... that's No Good (TM)
03:08 < valentin> just thought it would be good to get rid of the many stages concept as a far goal
03:08 < valentin> just distinguish beetween a basic toolchain and targets
03:08 < valentin> and the rest will be resolved automaticaly
03:08 * jsaw wished there would only be named stages
03:09 < jsaw> sth. like pre-cross, cross-toolchain, cli, gui, rebuild
03:09 < valentin> yea
03:10 < jsaw> I mean, I've never seen other stages than 012 3 5 and 7 (beside rebuild 9) used.. so what?
03:10 < valentin> full ack
03:11 < valentin> but that will be a todo for a rewritten build system
03:11 < valentin> i do not want to hack that into our current system
03:11 < jsaw> yeah I know. Just trying to get a better picture of the roadmap.
03:12 < valentin> another bonus would be, that we are able to distingish between packages, that are needed to build another package and resources that are possibly needed for the execution
03:13 < jsaw> ?
03:14 < valentin> we want to modularize the targets
03:14 < jsaw> how do u define "difference between program and resource"
03:15 < valentin> look at gcc - almost no program needs gcc to be executed, so you can have an installabe system that does not contain gcc
03:15 < valentin> but you have to build gcc for your toolchain
03:16 < jsaw> ah. okay. runtime and build dependency I would've called it...
03:16 < jsaw> yes, that'd be great.
03:17 < jsaw> but I fear this needs manual intervention.
03:17 < valentin> we want to get away from this one-sed-expression package selection
03:17 < valentin> sometimes, yes
03:18 < valentin> if you have a program that needs gcc, the user will have to put gcc into his target, too
03:18 < valentin> and there will be superfluos resources like build helper programs that are not marked as toolchain element
03:23 < mnemoc> jsaw: your mail about subconifg-init got 7.8 point on spamassasin :)
03:23 < mnemoc> X-Spam-Status: Yes, hits=7.8 required=6.0 tests=FROM_HAS_MIXED_NUMS,
03:23 < mnemoc> FROM_HAS_MIXED_NUMS3,RCVD_IN_DYNABLOCK,RCVD_IN_NJABL,RCVD_IN_SORBS,
03:23 < mnemoc> RCVD_IN_SORBS_WEB autolearn=no version=2.61
03:24 < valentin> which mail about subconfig-init ?
03:25 < valentin> oh, got it :_
03:25 < mnemoc> [t2] [RFC] subconfig-init.in (was Re: [t2-svn] rev 4671 - trunk/package/base/sysvinit)
03:25 < jsaw> mnemoc: ups
03:26 < valentin> jsaw: we often have things like that:
03:27 < valentin> package X and package Y are both build in two stages
03:27 < valentin> and they cross depend on each other
03:28 < valentin> so just skipping previous stages would not help to resolve that
03:28 < mnemoc> valentin: which packages are those?
03:30 < valentin> many basic packages build in stages 0,1,2
03:30 < valentin> of course there is no crossing if you take into account only one single stage
03:30 < valentin> but you cannot see that in the .cache files
03:31 < valentin> i just want to hack a program that reads bothe cache files and .d tags and trys to uncross those packages
03:33 < mnemoc> yeah, a good flist/dependencies analizer is needed
03:34 < valentin> binutils and gcc are the main trouble startes
03:34 < valentin> +r
03:34 < valentin> for they are build in all of the first stages
03:35 < valentin> and they cross depend
03:36 < valentin> another problem is that there are packages in the dep list, that are not build in the first two stages
03:36 < valentin> that does not matter because of different build modes
03:36 < valentin> but you cannot see that in the cache file either
03:36 < valentin> with just a bit more info in those files we would be fine
03:37 < jsaw> hmm.
03:38 < valentin> as jsaw said, there should be more semantic stages like toolchain, precross whatever
03:38 < valentin> and then we could store different deplists for different groups of stages
03:39 < valentin> transitive reduction running for 40 minutes now - very inefficient
03:41 < valentin> i guess i will code something smarter after sleep
03:44 < valentin> what dou you think: consider A in stage 2 and B in stage 1 and 2. now A depends on B. should A be build before or after B in stage 2 or does it not matter at all ?
03:44 < valentin> i would say "after B"
03:45 < valentin> by simply cutting off all packages of previos stages in the analysis we will lose that dep
03:47 < jsaw> it doesn't matter, but I'd make the rule "after dep in same stage"
03:48 < jsaw> that's what I meant before. If there's still an open dependency to a higher (or same stage), keep the dependency, otherwise simply drop it
03:48 < valentin> and how do we work with deps into higher stages ? mark as non-existent for that stage ?
03:49 < mnemoc> gn8, cu tomorrow
03:49 < jsaw> if the stages are ordered properly, that would refer to a "optional dependency", I'd say.
03:49 < valentin> mnemoc: ic i bore you
03:50 < valentin> yep.
03:50 < mnemoc> valentin: :p
03:50 < jsaw> gn8 mnemoc , sleep well
03:50 < valentin> mnemoc: sleep well
03:50 < mnemoc> valentin: no.... i'm just too tired..
03:50 < mnemoc> valentin: read my last posts to t2's ML to confirm that :p
03:53 < valentin> hehe
03:53 < valentin> http://www.boingboing.net/2004/11/04/my_modest_proposal_t.html
03:53 * valentin crawls soon into bed
03:54 < valentin> i'll code a dependency analyser that will order the graph using stages and then we'll see where the problematic cases will be
03:56 < jsaw> ohoh, alread 4am... you are right, I should also crawl towards (...) and into my bed
04:07 * valentin too
04:09 < valentin> gn8
04:09 < jsaw> gn8 valentin
08:00 -!- You're now known as rxr
08:15 < rxr> oh - much to scrollback ... ;-)
08:55 < praenti> hi rxr
08:55 < praenti> rxr: perhaps you can help me with my ibook i get from hannes a few days ago. i don#t get a usable x display even x starts
08:56 < praenti> its a ibook g4 12"
09:18 < rxr> is the display flickering?
09:18 < rxr> the use the UseFBDev option in the driver's section to let the framebuffer change the mode ...
09:18 < rxr> I have some patches to fix X, but they are not in ROCK and not fully polished ...
09:19 < praenti> rxr: no. the display is complete black
09:19 < praenti> also a kill of the X server does not restore the display, so the only way for me is a reboot :-/
09:19 < praenti> and the only iso i found was your drock image
09:20 < praenti> and i thought you are working with that
09:20 < praenti> or is there an error in?
09:26 < praenti> tried these two configs: http://www.desertsol.com/~kevin/ppc/XF86Config-4.3.0
09:26 < praenti> http://geekounet.org/powerbook/files/XF86Config-4-g4
09:28 < praenti> also with a standard xf86config run
09:28 * praenti is perhaps too stupid to configure an ibook...
09:28 < praenti> i must leave now. have to work...
09:28 < praenti> cu
09:38 < rxr> re
09:38 < rxr> praenti: sorry I read scrollback ..
09:39 < rxr> hm - the "drock" ISOs work great ...
09:39 < rxr> maybe use UseFBDev ...
09:40 < rxr> those X configs are full of junk
09:40 < rxr> use the T2 example (ok the rock would do, too) and just set the video chip in the driver section and uncomment the UseFBDev option ..
09:57 < rxr> I just update the radeon patch in t2, and report here if the work benh did does finalize my work
11:20 < CIA-8> rene * r4672 /trunk/package/audio/gsmp/ (gsmp.conf gsmp.desc):
11:20 < CIA-8> * updated gsmp (0.0.5 -> 0.0.6) - custom release for t2, to fix
11:20 < CIA-8> gcc >= 3.4.x regression ;-)
11:21 < rxr> ^- jsaw I know I'm crazzy ... - should reduce our error count by one ;-)!
11:38 -!- tschmidt__ [~tschmidt@p508EB06E.dip.t-dialin.net] has quit ["Verlassend"]
11:55 < mnemoc> moin
11:57 < rxr> moin mnemoc
11:57 < mnemoc> moin rxr
12:08 < rxr> 958 builds total, 541 completed fine, 28 with errors.
12:08 < rxr> ^- sparc
12:08 < rxr> too bad all the qt based stuff will fail due to qt seg-faulting ... :-(((
12:12 < CIA-8> jsaw * r4673 /trunk/package/graphic/imagemagick/imagemagick.desc: * update imagemagick (6.0.7-1 -> 6.1.3-7)
12:13 < mnemoc> rxr: same with older QTs?
12:15 < rxr> not tried yet
12:15 < rxr> most likely a gcc on sparc compiler bug anyway ...
12:15 < jsaw> damn, imagemagick update creates shared files with libtool... fix comes in a few seconds...
12:16 < rxr> which files?
12:16 < jsaw> note to self: next time wait until the checks finish
12:16 < jsaw> libltdl.so is installed
12:16 < jsaw> there's an option "--enable-ltdl-install" ...
12:16 < jsaw> that's is Stupid (TM)
12:16 < jsaw> s/is//
12:17 < jsaw> hi all, btw.
12:17 < rxr> yeah, moin jsaw ! ;-)
12:17 < rxr> jsaw: can you check if svn up will fix the gsmp regression?
12:18 < jsaw> svn up running..
12:18 < rxr> haha. "/.": "Microsoft is opting not to support Itanium on its coming release of Windows Server 2003 Compute Cluster Edition. Instead, Microsoft will focus on AMD's offerings and Xeon."
12:19 < jsaw> hey Intel, in germany one would say "Eigentor"
12:20 < martin_> moin
12:20 < rxr> hm - /me wonders whether I should phone intel and ask for intaniums for t2 so we can fill that hole ;-)
12:20 < rxr> hi mnemoc
12:21 < rxr> hey martin_ even ...
12:21 [Users #t2]
12:21 [ _Ragnar_] [ daja77] [ martin_] [ praenti] [ sparc-kly]
12:21 [ CIA-8 ] [ jsaw ] [ mnemoc ] [ rxr ] [ valentin ]
12:21 -!- Irssi: #t2: Total of 10 nicks [0 ops, 0 halfops, 0 voices, 10 normal]
12:22 < mnemoc> hi rxr :)
12:22 < mnemoc> hi martin_
12:23 < rxr> martin_: @uni or @home? - /me soon off to uni, soo ...
12:23 -!- sparc-kly [~sparky@65-23-193-94.prtc.net] has quit ["Leaving"]
12:24 < jsaw> rxr: do u think updating gcc will be a problem in the middle of a build?
12:39 < rxr> nope - minor version shouldn't be ...
12:46 < jsaw> gotta go, cu later!
12:46 < martin_> hi rxr, I'm at uni ...
12:48 < rxr> me soon too
15:24 < praenti> rxr: looks like there is a problem with framebuffer. starting x reluslts in a black display but with VGA extension on a external monitor I see my X
15:27 < mnemoc> what's the way to load a parse-config before the others?
15:30 -!- nzg [~tschmidt@pD95F8D06.dip.t-dialin.net] has joined #t2
15:41 -!- nzg [~tschmidt@pD95F8D06.dip.t-dialin.net] has quit ["Verlassend"]
15:43 -!- nzg [~tschmidt@pD95F8D06.dip.t-dialin.net] has joined #t2
15:50 < CIA-8> mnemoc * r4674 /trunk/package/base/sysvinit/ (config.in parse-config subconfig-init.in sysvinit.conf):
15:50 < CIA-8> * INIT selectiong improvements continued...
15:50 < CIA-8> * ROCKCFG_PKG_SYSVINIT_INITSTYLE renamed to ROCKCFG_INIT_STYLE
15:50 < CIA-8> * stage-9 init detection, a better on-config default init auto-selection
15:50 < CIA-8> * add some comments and decouple .init/m4 routines from sysvinit's legacy styles
16:19 -!- mtr [~michael@Ha9a0.h.pppool.de] has joined #t2
16:33 -!- sparc-kly [~sparky@65-23-193-94.prtc.net] has joined #t2
17:04 < CIA-8> mnemoc * r4675 /trunk/scripts/Build-Tools:
17:04 < CIA-8> * Built-Tools changed to only use parse-config and parse-config-[0-9]
17:04 < CIA-8> * parse-config list sorted considering parse-config=parse-config-5
17:13 < CIA-8> mnemoc * r4676 /trunk/package/base/sysvinit/ (parse-config parse-config-0): * sysvinit's parse-config splitted to have different priorities
17:23 < valentin> mnemoc: here ?
17:25 < mnemoc> yes
17:25 < valentin> nice
17:25 < mnemoc> tell me
17:25 < valentin> still evaluating your patch
17:26 < mnemoc> which one?
17:26 < valentin> the parse config stuff
17:26 < mnemoc> do you see any side efect?
17:35 < valentin> not realy - still matches all the parseconfigs and i do not know about ordering side effects
17:35 < valentin> i am just wondering why you need the 4 lines
17:35 < mnemoc> i also removed that -x and *~ stuff
17:35 < valentin> i saw that
17:36 < valentin> you do not need that anymore
17:36 < mnemoc> parse-config-[0-9] match only the active files
17:36 < valentin> ok, now i understood the expression
17:36 < mnemoc> i grab the number, sort using it and replace it with the source command
17:36 < valentin> i do not know if it is specified that you can have sth like package-config-somealphanummericname
17:37 < mnemoc> there is non currently
17:38 < valentin> ok, then we just say there must not be one in the future :)
17:39 < mnemoc> =)
17:41 < rxr> re
17:41 < valentin> hi
17:41 < rxr> anyone with domain right knowledge ...?
17:41 < rxr> s/right/law/ ...
17:43 < rxr> just to have some technically skilled "zeugen" could you look up rebe.de and confirm that it is not used?
17:44 < rxr> although valentin already known I monitor it for 4 years or so now ...
17:48 < rxr> valentin: we should sell technology to "dpinfo.dpma.de" so replace the existing crappy server either down or timeouting ...
18:12 -!- nzg [~tschmidt@pD95F8D06.dip.t-dialin.net] has quit ["Verlassend"]
18:20 < CIA-8> mnemoc * r4677 /trunk/package/contrib/directfb/directfb.desc: * directfb, no checksum for cvs:// yet
18:29 -!- martin_ [~martin@brln-d9ba255a.pool.mediaWays.net] has quit [Read error: 110 (Connection timed out)]
18:46 < CIA-8> mnemoc * r4678 /trunk/package/base/runit/ (parse-config runit.conf): * adds support for $ROCKCFG_INIT to runit
19:53 < rxr> November 9, 2004: GDB 6.3 Released!
19:53 * rxr testcompiling on powerpc
19:54 < mnemoc> =)
19:55 < rxr> btw. martin was so crude to install gentoo on a via board since yesterday evening or so ... :-(
19:56 < mnemoc> RIP
19:56 -!- martin___ [~martin@brln-d9ba02eb.pool.mediaWays.net] has joined #t2
19:58 < martin___> rxr I will come around at 8:00 o'clock. It is okay for you? :-)
19:58 < rxr> 19:55 < rxr> btw. martin was so crude to install gentoo on a via board since
19:58 < rxr> yesterday evening or so ... :-(
19:58 < rxr> 19:56 < mnemoc> RIP
19:59 < rxr> hi martin - just when we talk about you *g*
19:59 < mnemoc> =)
19:59 < rxr> but now martin happily want's to help improving our documentation the only part we are behind gentoo *g*
20:00 < mnemoc> =)
20:00 < martin___> ups. yes, gentoo is little strange. but it will work
20:00 < rxr> martin___: well that is quite early - but since I want to get rid of my empty AT-powered case and otherwise would need to carry it myself to the uni, ... come around ;-)
20:01 < rxr> should I take my sun with me and clone the system onto another U5 there so noone can take it anymore?
20:01 < rxr> martin___: will or might work? *g*
20:01 < rxr> btw, gentoo is some stange source distribution that is quite slow to compile on each system and not the filemanager you might have in mind
20:01 < mnemoc> can :)
20:03 < rxr> http://freshmeat.net/projects/gentoo/
20:03 < martin___> $ emerge gnome, and gnome will build and install very easy and with no problems.
20:03 < rxr> http://www.obsession.se/gentoo/
20:04 < rxr> http://www.gentoogames.com/
20:04 < rxr> ^- oh no - that is not the main site *g* - this is some stange shooter game they market or so
20:05 * rxr puzzled ...
20:05 < mnemoc> if my favorite tools is 'find' i should call my next distro with that name
20:05 < rxr> that is the site: http://www.gentoo.org/
20:05 < rxr> hard to find with M$ high tech search engine ;-)
20:06 < rxr> http://search.msn.com/
20:06 < rxr> so enought links to increase our google rating ;-)
20:06 < rxr> enough even ...
20:06 < mnemoc> =)
20:07 < rxr> == 20:05:07 =[9]=> Building develop/gdb [6.3 2.1.0-beta].
20:07 < rxr> ^- living here: http://www.gnu.org/software/gdb/gdb.html
20:07 < rxr> *g*
20:09 -!- martin___ [~martin@brln-d9ba02eb.pool.mediaWays.net] has quit ["Oops ..."]
20:11 < mnemoc> rxr: what do i need to svn emerges with http/https support?
20:12 < rxr> in gentoo or in t2 ?
20:12 < rxr> or this other evel thing with the name we do not want to mention *g*
20:13 < rxr> for http it needs neon
20:13 -!- martin___ [~martin@brln-d9ba02eb.pool.mediaWays.net] has joined #t2
20:13 < martin___> re
20:14 < rxr> hi mnemoc
20:14 < rxr> martin___: even ...
20:14 < rxr> damn - one of needs a new nick ...
20:14 * rxr should afford a "hi"-bot ...
20:15 -!- martin___ is now known as _martin
20:15 < rxr> doesn't help ...
20:15 < _martin> ?
20:16 < rxr> my client is inteligent enough to strip non ascii chars on completion ...
20:16 < _martin> :-(
20:16 * rxr could just train to hit the right keys ... ;-)
20:16 < rxr> _martin: renaming to gentooer would help ;-)
20:16 < rxr> just a joke ;-)
20:17 < mnemoc> rxr: t2.... was neon included inside svn?
20:17 < mnemoc> wasn't*
20:18 < rxr> yes _was_ no idea if it still is ...
20:20 < _martin> So, I will genhaehse a new green cat. Or a reindeer maybe ... http://kif.fsinf.de/drachen/katzen.html, http://kif.fsinf.de/drachen/rentier.html
20:21 < rxr> ?
20:22 < rxr> jsaw: did you test the gsmp update?
20:43 < jsaw> re
20:43 < jsaw> hi all
20:43 < jsaw> rxr: yes, compiles.
20:44 < rxr> cool
20:44 < rxr> do not try to use it ...
20:44 < rxr> ;-(
20:44 * rxr working on the usage part ...
20:45 < jsaw> but the imagemagick update is annoying me
20:46 < jsaw> /usr/lib/gcc/i586-unknown-linux-gnu/3.4.2/libstdc++.la: No such file or di
20:46 < jsaw> rectory
20:46 < jsaw> ???
20:46 < jsaw> wtf is python hardcoding this into its config???
20:47 < mnemoc> o_O
20:48 < jsaw> (well, I think it's python, actually kdebindings is failing due to this)
20:51 < rxr> hm - for gdb the patch needs to be updated, and it has regressions on powerpc :-(
20:51 < rxr> if someone else want's to update it ...
20:52 < jsaw> which patch?
20:53 < rxr> the installed-by-binutils.patch
21:03 < jsaw> guess, we should install the default libstdc++ into /usr/lib...
21:04 < jsaw> (only, I mean)
22:41 < rxr> 22:21 < rxr> 958 builds total, 578 completed fine, 32 with errors.
22:41 < rxr> 22:21 < rxr> ^- sparc32
22:41 < rxr> 22:22 < rxr> btw nx compressino stuff got an update, packaging anyone?
22:41 < rxr> ^- was on the wrong channel ...
22:41 < mnemoc> :D
22:41 < rxr> .oO (when gsmp-0.0 is run in valgrind on PowerPC the threading stuff does not work )
22:42 < rxr> it produces totally randomm stuff ... that is not in the exacution path ...
22:42 < mnemoc> o_O
22:43 < rxr> the valgrind ppc version is also not yet int he official valgrind tree - still quite very alpha ...
22:43 < mnemoc> ic
22:43 < rxr> from the results, ic, too
22:44 < mnemoc> there can be any difference in valgrind's output runing on a different arch?
22:45 < rxr> well - since valgrind visualizes and dynamically translates the instructions to some extend it is highly CPU depenant ...
22:45 < rxr> at first it only worked on x86, the ppc stuff is bleeding edge ..
22:46 < mnemoc> but a memory problem is on your c code, then that doesn't change if you use ppc or x86 instructions
22:46 < rxr> well - if it is random corruption, it will happen at a totally different place
22:47 < mnemoc> random corruption?
22:47 < rxr> and the other thing is that the ppc valgrind does not yet work reliable ...
22:47 < mnemoc> that's my point
22:47 < rxr> not to mention that it does not work with threaded apps at all ...
22:47 < rxr> (at least on ppc)
22:48 < mnemoc> if valgrind/gsmp pass on x86, you can assume it will work on any other arch
22:48 < rxr> ouhm - I would dream about that, too
22:48 < mnemoc> :\
22:48 < rxr> I debugged far too many non-x86 arch problems ...
22:48 < rxr> endianess, va_list differnces ...
22:49 < mnemoc> va_list?
22:49 < rxr> size of *int*, some default to signed - some to unsigned ...
22:49 < rxr> man va_start ...
22:49 < rxr> never used such?
22:49 < mnemoc> yes
22:50 < mnemoc> i don't understand how va_list can act different in a different arch
22:50 < rxr> An obvious implementation would have a va_list a pointer to the stack
22:50 < rxr> frame of the variadic function. In such a setup (by far the most com-
22:50 < rxr> mon) there seems nothing against an assignment
22:50 < rxr> va_list aq = ap;
22:50 < rxr> Unfortunately, there are also systems that make it an array of pointers
22:50 < rxr> (of length 1), and there one needs
22:50 < rxr> va_list aq;
22:50 < rxr> *aq = *ap;
22:50 < rxr> Finally, on systems where parameters are passed in registers, it may be
22:50 < rxr> necessary for va_start to allocate memory, store the parameters there,
22:50 < rxr> and also an indication of which parameter is next, so that va_arg can
22:50 < rxr> step through the list. Now va_end can free the allocated memory again.
22:50 < rxr> To accommodate this situation, C99 adds a macro va_copy.
22:50 < rxr> ^- and have fun ..
22:50 < mnemoc> oh
22:51 < rxr> mnemoc: you only have x86 over there, hm? ;)
22:51 < mnemoc> isn't it obvious? :(
22:51 < rxr> signedness of char is also a nice game most people are not aware off ... :-(((
22:51 < rxr> mnemoc: it serves your nerves ;-)
22:51 < mnemoc> he
22:52 < rxr> saves your nerves even ...
22:52 < rxr> or safe ....
22:52 < mnemoc> )
22:52 < mnemoc> :)
22:52 * rxr too puzzled from debugging seg-faults from a two year old peace of code ...
22:52 < rxr> (now guess what ;-)
22:53 < rxr> and I think I stop that now ...
22:53 < jsaw> http://www.mini-itx.com/projects/cluster/
22:53 < mnemoc> :)
22:53 * rxr should spend time developing on the next generation code than debugging this s**^stuff
22:53 < valentin> re
22:53 < valentin> rxr: yep
22:54 < rxr> #6 0x0fd69954 in GSMP::App::CreateMainWindow (this=0x0, view_type=@0xffffffff)
22:54 < rxr> at stl_deque.h:946
22:54 < rxr> ?!?!?
22:56 < valentin> rxr: i guess you'll never get that running :/
22:56 < valentin> does it still segfault in some dl loading code ?
22:57 < valentin> btw: we shoulkd migrate that discussion to #gsmp
22:58 < rxr> no - currently at:
22:58 < rxr> #0 0x0f0f5dfc in StatusBar::StatusBar ()
22:58 < rxr> from /usr/local/lib/gsmp/GUI-gtk.plugin.so
22:58 < rxr> but my feeling is that the loaded .so's do not work reliable ...
22:59 < rxr> the guy uses an ugly netgear switch - bad choice ... - I never buy such crap again ...
23:02 < rxr> ok - me evening lunch - and then with a book into bed ...
23:05 < rxr> valentin: don't say never - this just motivates me to _get_ it going ...
23:05 < rxr> _martin: the via link above is for you ...
23:09 < CIA-8> jsaw * r4679 /trunk/package/graphic/imagemagick/ (imagemagick.conf no-ltdl-install.patch):
23:09 < CIA-8> * imagemagick: get rid of the shared libltdl.so* files with libtool
23:09 < CIA-8> (thanks to a guy name Jakub...)
23:09 < CIA-8> * cleanup conf
23:27 < rxr> Jakub?
23:29 < jsaw> http://www.mail-archive.com/pld-devel-en@pld-linux.org/msg00836.html
23:30 < mnemoc> i knew this will happen.... th changed the prefix of svn on 2.0-stable, and my svn server crashed because command line was 1.0.8 and mod_svn 1.1 .... apache 100% :|
23:35 -!- sparc-kly [~sparky@65-23-193-94.prtc.net] has quit [Read error: 110 (Connection timed out)]
23:36 < mnemoc> http://www.php.net/~sterling/mono/ <--- this is weird
--- Log closed Sat Nov 13 00:00:15 2004