| 00:00.47 | mumptai | well, i think i have enough max232s at home ... therefore :) |
| 00:00.59 | mumptai | okay, i think i'll try it |
| 00:01.15 | mumptai | thx |
| 00:02.20 | phlog-mobile | how much of the OS is in the flash card? Does any of it get put in the drives you format with its software? |
| 00:04.33 | [g2] | jacques, howdy! |
| 00:04.45 | jacques | phlog-mobile, it might help if you specify stock firmware or unslung 1/2/3 |
| 00:04.51 | jacques | hi [g2] how's it going? |
| 00:05.10 | [g2] | good, back from holiday break |
| 00:05.22 | jacques | cool. seems quiet here (understandably) |
| 00:05.29 | [g2] | ya |
| 00:05.40 | [g2] | did you see that TMPDIR patch the other day ? |
| 00:05.53 | jacques | no I missed that - what is it? |
| 00:06.16 | phlog-mobile | jacques, ah. How about Unslung 3? |
| 00:06.22 | phlog-mobile | vs. stock |
| 00:06.57 | [g2] | jacques, TMPDIR = "${TOPDIR}/tmp-${TARGET_OS}" |
| 00:07.07 | *** join/#nslu2-linux McTails (~mcasadeva@ool-182d2d1c.dyn.optonline.net) |
| 00:07.12 | jacques | stock runs from ramdisk and unslung3 runs from internal flash |
| 00:08.09 | jacques | [g2], ah I saw you discussing that I recall - what does that get you? |
| 00:08.16 | [g2] | jacques, I've got both tmp-linux and tmp-linux-uclibc dirs |
| 00:08.27 | [g2] | I can just change TARGET_OS and have both |
| 00:09.00 | [g2] | a libc and uclibc version |
| 00:09.17 | jacques | ah OK. I just have multiple build dirs |
| 00:09.23 | [g2] | exactly :) |
| 00:09.33 | [g2] | same sw base |
| 00:09.44 | [g2] | literally same :) |
| 00:10.05 | [g2] | well, except for tool chain diff |
| 00:10.06 | [g2] | s |
| 00:11.06 | [g2] | jacques, I'd like to clean up the startup a little and have an openslug appliance build |
| 00:11.12 | jacques | I've been kinda stuck lately at several fronts - the hh.o being down has made it impossible to research the ipkg/uclibc problem, openslug-glibc seems to have endian issues, and I need rwhitby-away to help me test the new slugimage or change the build system to use a cvs data for slugimage before committing the new one |
| 00:11.49 | jacques | (can't have all builds suddenly using the new slugimage when it's not validated) |
| 00:11.54 | [g2] | I'm running python with a web server, lsof, strace, and bittorrent tracker on my openslug |
| 00:12.01 | jacques | glibc? |
| 00:12.05 | [g2] | yep |
| 00:12.10 | jacques | what happens when you run top ? |
| 00:12.26 | jacques | see things like this?: |
| 00:12.29 | jacques | Load average: 0.00, -45802652779840844014642954932011411854766603428837684131471 |
| 00:12.29 | jacques | 07019245672921788002139630714066970149111054132292212095975424.00, -0.00 |
| 00:12.46 | jacques | what about if you were to build a wget package (instead of the bb one) |
| 00:13.10 | jacques | see things like this: |
| 00:13.12 | jacques | <PROTECTED> |
| 00:13.12 | jacques | 06083122925498778585131084098570036107192739755968101053519794064067384265220017348507488976834896963172613459113069300929974 |
| 00:13.12 | jacques | 869937881088.00 MB/s |
| 00:13.36 | [g2] | there was a libcurses issue when I install the curses python pkg |
| 00:13.43 | jacques | or this: wget: progress.c:830: create_image: Assertion `p - bp->buffer <= bp->width' failed. |
| 00:14.09 | [g2] | wget's work fine for me |
| 00:14.19 | jacques | wget 1.9.1 > NOT the bb wget ? |
| 00:14.26 | [g2] | I've done 40+ to install the full python |
| 00:14.35 | [g2] | ah |
| 00:14.41 | jacques | <jacques> what about if you were to build a wget package (instead of the bb one) |
| 00:15.23 | jacques | I thought this was a soft-float issue but I build a hard-float version (at least I asked oe to) and it behaves the same |
| 00:15.40 | jacques | so now it seems like a glibc issue |
| 00:17.01 | [g2] | top load average looks bad, but that's the only thing in top (during a 5 second glance) |
| 00:17.36 | jacques | yeah same here |
| 00:17.54 | [g2] | did you install the ixp425 sw |
| 00:18.44 | jacques | the what? |
| 00:19.00 | [g2] | modprobe ixp425_eth |
| 00:19.02 | kergoth | jacques: those number problems could be a locale problem... |
| 00:19.04 | jacques | oh yep |
| 00:19.11 | jacques | (to [g2]) |
| 00:19.35 | [g2] | root@nslu2:~# date |
| 00:19.36 | [g2] | Mon Dec 27 08:23:26 UTC 2004 |
| 00:19.36 | [g2] | root@nslu2:~# |
| 00:19.45 | jacques | kergoth well I am building with USE_NLS = "no" |
| 00:20.00 | kergoth | ah, good |
| 00:20.13 | jacques | not good because it means no easy fix :-( |
| 00:20.16 | [g2] | jacques, I did too, but the glibc build ignored that |
| 00:20.28 | jacques | [g2], argh! really? |
| 00:20.49 | [g2] | so I don't think it's flowing all properly through oe |
| 00:21.09 | [g2] | there's like 200 ipkgs for all the locales stuff |
| 00:21.23 | kergoth | if you were to disable locale support in glibc completely, you'd break binary compatibility, which is why it doesnt obey USE_NLS atm |
| 00:21.40 | [g2] | kergoth, that makes sense |
| 00:21.47 | kergoth | in fact, there's no actual way to disable it in glibc's buildsystem. |
| 00:21.56 | kergoth | i debated adding support for such a thing, but its debatable |
| 00:21.59 | kergoth | heh |
| 00:22.00 | [g2] | that makes some sense too |
| 00:22.03 | kergoth | so didnt bother |
| 00:22.08 | jacques | export LANG=C still does it |
| 00:22.33 | jacques | as does En_US |
| 00:22.33 | [g2] | I'm just wondering if something gets set there and flows through afterwards |
| 00:23.03 | jacques | kergoth any suggestions for me to try? |
| 00:23.06 | kergoth | well, you probably dont have the locale files installed for en_US. |
| 00:23.08 | kergoth | heh |
| 00:23.18 | [g2] | I've got the ipkg :) |
| 00:23.30 | jacques | yeah but C ... |
| 00:23.40 | kergoth | right, C should work. dunno. |
| 00:23.51 | kergoth | could be that it isnt a locale problem, just a suggestion |
| 00:24.04 | [g2] | we'll I should be able to install the locale file rigth ? |
| 00:24.12 | [g2] | then the problem should stop |
| 00:24.32 | kergoth | we havent yet determined that its a locale issue. and jacques just tested with LANG=C, which shouldve have worked. |
| 00:24.41 | [g2] | or is it a miss config between busybox and glibc |
| 00:24.46 | jacques | I'm a natural pessimist I admit, but I don;t think it's going to me nearly that simple |
| 00:25.15 | jacques | if using a more recent glibc wasn't so black magicky I would have tried it already |
| 00:25.42 | [g2] | I'm running 2.3.3 iirc |
| 00:25.49 | kergoth | hm? you can use current cvs or 2.3.3 easily enough. |
| 00:26.01 | kergoth | just a matter of setting PREFERRED_VERSION_glibc |
| 00:26.28 | [g2] | I"m running 2.3.3 |
| 00:27.06 | [g2] | root@nslu2:/lib# ls -l libc.so.6 |
| 00:27.06 | [g2] | lrwxrwxrwx 1 root root 13 Jan 1 1970 libc.so.6 -> libc-2.3.3.so |
| 00:27.12 | jacques | kergoth oops I missed the glibc_cvs.oe - sorry |
| 00:27.21 | *** join/#nslu2-linux sandos (HydraIRC@130.243.93.185) |
| 00:27.44 | kergoth | jacques: that fixed date 2.3.2 is just cvs locked at an arbitrary date. |
| 00:27.48 | kergoth | we used to use _cvs by default |
| 00:27.49 | kergoth | heh |
| 00:27.57 | jacques | hmm I think I'm running glibc_2.3.2+cvs20040726.oe |
| 00:28.18 | jacques | but if [g2] is running 2.3.3 and still has the issue, I should probably try cvs |
| 00:28.20 | [g2] | kergoth, btw congrats on the Dell monitor, I'm happy for your |
| 00:28.24 | kergoth | 2.3.2 as released is useless |
| 00:28.42 | kergoth | too many bugs, even debian uses cvs |
| 00:28.54 | [g2] | hows 2.3.3 ? |
| 00:29.04 | kergoth | heh, thanks, still need to get it exchanged though. there's a small scratch on the lcd screen :( |
| 00:29.09 | kergoth | dunno |
| 00:29.13 | kergoth | havent used it |
| 00:29.16 | [g2] | :( |
| 00:29.41 | jacques | [g2], did you specify 2.3.3 in your build? because glibc_2.3.2+cvs20040726.oe id's as 2.3.3 |
| 00:29.59 | jacques | libc-2.3.3.so - I have that too |
| 00:30.13 | kergoth | right, cvs, so it has 2.3.3 symbols |
| 00:30.23 | [g2] | jacques, I didn't specify any thing other than TARGET_OS = "linux" |
| 00:30.41 | kergoth | [g2]: then it used 2.3.2+cvs20040726. |
| 00:30.43 | jacques | [g2], OK then we are running the same version |
| 00:30.44 | kergoth | thats default for most platforms |
| 00:30.55 | jacques | I'll try some newer ones |
| 00:31.15 | [g2] | as an example glibc-localedata-fr-lu_2.3.2+cvs20040726-r11_armeb.ipk |
| 00:32.20 | [g2] | jacques, kergoth python was running a simple http server with glibc |
| 00:32.35 | kergoth | cool |
| 00:32.46 | [g2] | and the bittorrent tracker |
| 00:33.52 | [g2] | kergoth is there an easy way to make a native dev system ? |
| 00:34.11 | *** part/#nslu2-linux sandos (HydraIRC@130.243.93.185) |
| 00:34.16 | [g2] | bye |
| 00:34.21 | jacques | glibc based should work now according to kergoth :-) |
| 00:35.13 | kergoth | yep, should work fine |
| 00:35.16 | [g2] | re "<kergoth> thats default for most platforms" |
| 00:35.31 | kergoth | ? |
| 00:36.12 | [g2] | the glibc we using 2.3.2+cvs20040726 is ok right ? |
| 00:36.19 | [g2] | s/we/we're/ |
| 00:36.28 | kergoth | it wouldnt be default if it wasnt. |
| 00:36.51 | kergoth | i fixed native dev a while ago |
| 00:36.54 | kergoth | jacques: uclibc works too. |
| 00:37.05 | jacques | kergoth wow really? excellent! |
| 00:37.07 | [g2] | ok so back to my other question: native dev ? |
| 00:37.15 | kergoth | ? |
| 00:37.18 | jacques | lol |
| 00:37.23 | kergoth | are you blind, or just lagged? |
| 00:37.28 | jacques | [g2], that's what we were just talking about |
| 00:37.42 | kergoth | 18:25 < jacques> glibc based should work now according to kergoth :-) |
| 00:37.45 | kergoth | 18:27 < kergoth> i fixed native dev a while ago |
| 00:37.58 | jacques | oemake gcc |
| 00:37.59 | [g2] | I saw those |
| 00:38.14 | kergoth | so install gcc, binutils, and uclibc-dev or libc6-dev and go compile away. |
| 00:38.40 | [g2] | right. |
| 00:39.08 | [g2] | I was wondering if there was a grouping of packages like gcc, binutils, auto*, etc.... |
| 00:39.09 | jacques | the talk of python reminded me that I have been intending to contribute the data I collected from my native builds of perl to oe so it can cross-build it properly |
| 00:39.15 | kergoth | no. |
| 00:39.23 | kergoth | install what you need. |
| 00:39.38 | kergoth | i havent created a meta package for C development yet |
| 00:39.49 | jacques | I did in sept |
| 00:40.01 | jacques | that's when I discovered gcc wasn't quite working |
| 00:40.07 | kergoth | jacques: noidd added the crosscompile bits to perl and got them into perl cvs as a part of his work to make it crosscompile for the openzaurus buildsystem. |
| 00:40.10 | [g2] | jacques, care to share :) |
| 00:40.16 | kergoth | just talk to him about fixing the .oe |
| 00:40.18 | jacques | let me see if I can still find it :-\ |
| 00:40.35 | jacques | kergoth great thanks |
| 00:40.39 | kergoth | [g2]: without a perl ipk, you have no automake. |
| 00:41.03 | [g2] | I'm really trying to run the perl and python test suites |
| 00:41.14 | [g2] | and glibc too would be nice |
| 00:41.15 | jacques | trying to remember which machine i was on when I did it |
| 00:41.23 | jacques | yeah good luck with dejagnu |
| 00:41.39 | [g2] | jacques, I can just run them natively |
| 00:42.07 | jacques | you need dejagnu to run the glibc testsuite (at least you did last time I did it) |
| 00:42.13 | [g2] | well, it a week or two I'll be able to run them natively :) |
| 00:42.37 | jacques | I've run the perl tests when i did the native builds of it |
| 00:42.47 | [g2] | on the slug ? |
| 00:42.51 | [g2] | openslug ? |
| 00:42.58 | jacques | yes - I also did that in sept |
| 00:43.11 | jacques | using a buildroot-built dev env |
| 00:43.24 | [g2] | that was with uClibc |
| 00:43.48 | jacques | yes |
| 00:43.51 | [g2] | different gcc (probably 3.4) |
| 00:43.58 | [g2] | different binutils |
| 00:44.23 | kergoth | so what? that wont affect the ability to run the tests. jacques said he ran them, not htat they all passed. |
| 00:44.26 | kergoth | heh |
| 00:44.37 | [g2] | lol |
| 00:45.30 | jacques | (still looking for the devel env package) |
| 00:45.57 | [g2] | well, being the lazy bastard that I am, having the perl, python, glibc, ltp and possibly posix tasts already developed, I think it's at least worth looking at running those tests |
| 00:46.16 | [g2] | s/tasts/tests/ |
| 00:48.58 | jacques | [g2], let me know if you get the glibc tests going . I intend to never touch dejagnu again if I can avoid it |
| 00:50.03 | jacques | OK I found what I had done for devel - I copied task-boostrap to task-devel.oe and I added: |
| 00:50.17 | jacques | bash binutils coreutils diffutils gawk gcc grep make sed to DEPENDS |
| 00:50.21 | *** join/#nslu2-linux ByronT (~byrontodd@1Cust246.tnt1.gadsden.al.da.uu.net) |
| 00:50.36 | jacques | and bash binutils coreutils diffutils gawk gcc grep libc6-dev make sed to RDEPENDS |
| 00:52.50 | *** join/#nslu2-linux sandos (HydraIRC@130.243.93.185) |
| 00:53.54 | [g2] | jacques, thx |
| 00:54.09 | [g2] | jacques, what was the big issue with deja-gnu ? |
| 00:54.25 | jacques | it's software from hell |
| 00:54.49 | jacques | impossible to build, configure, install, and use IMHO |
| 00:54.51 | [g2] | badly written or complex |
| 00:55.33 | jacques | maybe you'll get along better with it |
| 00:55.59 | [g2] | kergoth know any glibc guys or gcc guys ? |
| 00:56.29 | [g2] | gals are fine too, that was the generic gals/guys |
| 00:57.33 | [g2] | i heard there were big changes in how the trees were represented in 3.5 i think |
| 00:58.13 | kergoth | yeah, i read a bit about that. dont know any, nope |
| 00:58.45 | kergoth | need to get ahold of a gcc person, theres what appears to be a flaw in the way the gcc buildsystem handles canadian cross |
| 00:59.02 | *** part/#nslu2-linux sandos (HydraIRC@130.243.93.185) |
| 00:59.15 | [g2] | the current gcc or the new one ? |
| 00:59.24 | [g2] | wb |
| 00:59.32 | kergoth | current |
| 00:59.41 | kergoth | had to hack around it to make the native toolchain work properly |
| 01:13.04 | *** join/#nslu2-linux christopher_logg (~christoph@plo-cr1-212-105.peak.org) |
| 01:19.13 | *** join/#nslu2-linux Christopher_bbl (~Christoph@plo-cr1-212-105.peak.org) |
| 01:23.12 | *** join/#nslu2-linux rwhitby-treo (upirc@203.202.55.207) |
| 01:24.06 | rwhitby-treo | ~botsnack turkey |
| 01:24.06 | jbot | thanks, rwhitby-treo |
| 01:25.44 | [g2] | is that a 650 yet ? |
| 01:25.59 | [g2] | happy thanksgiving :) |
| 01:27.13 | rwhitby-treo | thanksgiving? what's that? :-) |
| 01:27.32 | rwhitby-treo | treo650 GSM not released yet :-( |
| 01:28.08 | rwhitby-treo | Ksilebo has one |
| 01:28.25 | Ksilebo | Muahaha |
| 01:29.11 | Ksilebo | Though it is on Sprint PCS and I had to lease out my soul. |
| 01:29.46 | rwhitby-treo | I'm hoping that they fix the memory issue by putting 64Mb in the international GSM version (to fit in the extra languages) |
| 01:30.03 | Ksilebo | I dont know what they're going to do. |
| 01:30.40 | [g2] | <PROTECTED> |
| 01:30.41 | [g2] | <PROTECTED> |
| 01:30.41 | [g2] | <PROTECTED> |
| 01:30.41 | [g2] | <PROTECTED> |
| 01:30.42 | [g2] | <PROTECTED> |
| 01:30.50 | [g2] | who put that in ? |
| 01:30.50 | Ksilebo | They should have had 64+ MB to begin with. |
| 01:31.19 | Ksilebo | Thank god I had my 256 MB SD card from my 600. |
| 01:31.31 | rwhitby-treo | they'll placate people with free 128Mb SD cards, and then just half the sector size |
| 01:31.53 | rwhitby-treo | g2: that's standard OE |
| 01:32.14 | [g2] | rwhitby-treo, test is backwards right ? |
| 01:32.35 | [g2] | test -z is for zero lenght right ? |
| 01:32.37 | rwhitby-treo | dunno |
| 01:33.06 | [g2] | so If -z then fakeroot and explode on my system |
| 01:33.50 | rwhitby-treo | has it changed recently, cause we've been using that code for a while ... |
| 01:34.25 | rwhitby-treo | back later |
| 01:34.39 | [g2] | well just failed today |
| 01:35.02 | *** part/#nslu2-linux rwhitby-treo (upirc@203.202.55.207) |
| 01:41.49 | [g2] | kergoth ping |
| 01:42.56 | [g2] | jacques, ping |
| 01:45.36 | *** join/#nslu2-linux Christopher_bbl (~Christoph@plo-cr1-212-105.peak.org) |
| 01:45.40 | jacques | hi. sorry on phone |
| 01:45.56 | [g2] | hey np at all |
| 01:48.53 | Ksilebo | [g2]: Failed?! |
| 01:49.20 | [g2] | Ksilebo, Failed as in didn't build |
| 01:49.25 | Ksilebo | Oh |
| 01:49.38 | Ksilebo | I approve. Carry on. |
| 01:49.44 | [g2] | thx |
| 01:49.48 | [g2] | :) |
| 01:56.32 | *** join/#nslu2-linux Christopher_ (~Christoph@plo-cr1-212-105.peak.org) |
| 01:57.12 | jacques | hmm glibc-2.3.3 build failed :-( |
| 01:57.40 | [g2] | hmm oe acting funny :( |
| 02:02.45 | jacques | I wonder if I'm missing a patch or it's new binutils version or something |
| 02:19.41 | jacques | why does PREFERRED_VERSION_glibc = "cvs" not work? |
| 02:20.00 | jacques | it tried to build the default |
| 02:20.55 | [g2] | dunno, I'm chasing a autom4te issue for fakeroot-native not building anymore |
| 02:25.49 | kergoth | fakeroot-native ends up in your dependencies by default. |
| 02:25.53 | kergoth | jacques: uh? |
| 02:26.04 | kergoth | jacques: because 'cvs' isnt the version of that package, perhaps? |
| 02:26.28 | jacques | kergoth that's the easy one - hard one is why does glibc 2.3.3 build fail? |
| 02:26.50 | kergoth | how the fuck should i know? there's a reason we dont use it. |
| 02:27.02 | kergoth | it doesnt build for half the architectures thatve been tried |
| 02:27.15 | kergoth | heh |
| 02:27.40 | CIA-9 | 03rwhitby * 10unslung/sources/lsof/control: Bumped the version. |
| 02:27.49 | kergoth | 'PV' is a package's version. it doesnt have to be set in the filename, and even if it is, it can always be overridden in the .oe itself. go read glibc_cvs.oe. |
| 02:28.15 | CIA-9 | 03rwhitby * 10unslung/make/lsof.mk: Bumped the version. |
| 02:29.08 | jacques | but how do I get oe to even look at glibc_cvs.oe ? |
| 02:29.20 | kergoth | what are you talking about? |
| 02:29.39 | kergoth | i just told you. |
| 02:29.43 | kergoth | 'cvs' is not that package's version |
| 02:29.46 | kergoth | go look at what the version is |
| 02:29.52 | kergoth | and set that version in the preferred version variable |
| 02:29.55 | jacques | oh |
| 02:29.57 | kergoth | this isnt exactly rocket science |
| 02:30.20 | kergoth | PV = "2.3.2+cvs${CVSDATE}" |
| 02:30.35 | kergoth | PREFERRED_VERSION_glibc = "2.3.2+cvs${CVSDATE}" |
| 02:30.37 | kergoth | there, done. |
| 02:31.03 | jacques | PV = "2.3.2+cvs${CVSDATE}" |
| 02:31.14 | kergoth | yes, thats whats in the .oe. |
| 02:31.29 | jacques | looks like that's from before 2.3.3 |
| 02:31.50 | kergoth | yep |
| 02:31.58 | jacques | I guess the 2.3.2 is jmeaningless in this case ? |
| 02:32.10 | kergoth | meaningless? |
| 02:32.23 | jacques | (it's not going to actually start with 2.3.2 and then apply some patch) ? |
| 02:32.26 | rwhitby-away | glc: lsof is now -2 in the feed, and that should be an arm binary |
| 02:32.27 | kergoth | its the package version. it just happens that nobody updated it after 2.3.3 was tagged in cvs |
| 02:32.30 | kergoth | read SRC_URI |
| 02:32.34 | kergoth | SRC_URI is what sources are used |
| 02:32.38 | kergoth | PV has nothing to do with SRC_URI |
| 02:32.41 | kergoth | one is version, one is source |
| 02:32.42 | kergoth | go read. |
| 02:32.50 | kergoth | i really dont have time to hold your hand on this. try reading some .oe files. |
| 02:40.16 | jacques | I set PREFERRED_VERSION_glibc = "2.3.2+cvs20041125" and it's still trying to build the default |
| 02:40.43 | kergoth | of course it is. if CVSDATE != 20041125, what do you think is going to happen? |
| 02:41.02 | kergoth | if CVSDATE != 20041125, that isnt the version of the .oe, and there's no .oe in the metadata that provides the version you requested |
| 02:42.12 | jacques | interesting |
| 03:20.50 | *** join/#nslu2-linux NSLU2-IRC (~NSLU2-IRC@wbar8.sea1-4-4-087-012.sea1.dsl-verizon.net) |