**** BEGIN LOGGING AT Tue Feb 20 02:59:57 2007 Feb 20 03:15:08 Does GCC 4.1.2 have any worthwhile improvements for Arm processors over 4.1.1? Feb 20 03:15:26 I'm having a hard time justifying an upgrade to it. Feb 20 03:42:08 Running debian slug w/ kernel 2.6.18 but it has a bug with rtorrent, how can I upgrade to 2.6.20? Feb 20 04:22:39 AwayNAiL: make /home/slug/optware/ts101/toolchain/buildroot/.configured Feb 20 04:22:39 is not stable. Feb 20 04:22:49 (i.e. it always rebuilds) - that's the root of the problem. Feb 20 04:25:49 it should be moved outside of ifeq ts101 Feb 20 04:26:08 i mean "touch .../.configured" Feb 20 04:31:17 03bzhou * r5551 10optware/trunk/scripts/optware-check-package.pl: added target to bintype mapping for ts101 Feb 20 04:45:24 eno: I'm testing this with ts101, so that's not it Feb 20 04:46:00 ah, sorry - ifneq. Feb 20 04:46:41 yep, that would do it Feb 20 04:47:49 03rwhitby * r5552 10optware/trunk/make/buildroot.mk: Fixed .built target Feb 20 04:48:04 AwayNAiL: ^^^ try that. Feb 20 05:12:26 03bzhou * r5553 10optware/trunk/make/buildroot.mk: buildroot: more tailoring for ts101 Feb 20 05:20:42 03bzhou * r5554 10optware/trunk/make/libiconv.mk: libiconv: to build when libiconv in $(PACKAGES) Feb 20 05:33:21 03bzhou * r5555 10optware/trunk/Makefile: added TS101_ variables and toolchain Feb 20 05:49:37 rwhitby: we can try enabling ts101 on nudi now Feb 20 05:51:03 I'll kick off a test build first. Feb 20 06:53:48 03oleo * r5556 10optware/trunk/ (3 files in 2 dirs): transmission: r1504 - added one day offset to gnuplot graphs Feb 20 07:32:35 03oleo * r5557 10optware/trunk/make/firmware-oleg.mk: fimware-oleg: add PATH to uClibc build Feb 20 10:18:58 rwhitby: Have you done a test build? Feb 20 10:19:10 (And did it build actual ppc binaries?) Feb 20 10:21:35 NAiL: we fixed the rebuild problem (eno spotted it) Feb 20 10:21:43 I stopped after that. Feb 20 10:22:25 ah Feb 20 10:22:32 I'm building now anyway Feb 20 10:36:26 thanks eno/rwhitby :-) Feb 20 10:47:35 rwhitby: bingo. It works :-D Feb 20 10:47:48 It doesn't loop on the toolchain, nor does it build native binaries. Feb 20 10:48:00 Now it's time to see if it builds the whole feed :) Feb 20 10:48:16 NAiL: tip for the future, run each of the make targets twice (starting a .configured, .built, etc) and make sure that the second one does nothing. Feb 20 10:49:03 yeah Feb 20 10:51:31 cool, eno has added Oleg Feb 20 10:51:36 's firmware to Optware :-) Feb 20 10:53:28 mysql 404's Feb 20 10:53:44 it continues to build though Feb 20 12:17:31 rwhitby, NAiL ... can any of you tell me how i get the "make debianslug" commands to actually start building firmware again ... after a week or more of inactivity on my side i updated my source-tree only to discover the make command no longer function because of bitbake complaiing about "ERROR: You must set MACHINE to nslu2le or nslu2be." .... is this a mistake somewhere in the debianslug/openslug distro/*.conf files ... or do Feb 20 12:17:31 i need to actrually setup a local.conf file to specify this endianness ? Feb 20 12:18:02 Caelian|work: are you subscribed to the nslu2-linux mailing list? Feb 20 12:18:11 rwhitby: no i am not Feb 20 12:18:24 do you have a mailinglist archive link ? Feb 20 12:18:35 that's the first problem. How do you expect to find out about these things then? Feb 20 12:19:09 you need to edit conf/auto.conf Feb 20 12:19:15 DISTRO="slugos" Feb 20 12:19:24 MACHINE="nslu2be" or "nslu2le" Feb 20 12:19:48 or even better: Feb 20 12:19:57 mv debianslug debianslug_to_be_deleted Feb 20 12:20:00 make slugosle Feb 20 12:20:34 rwhitby: i guess i should probably add that subscription to my already 20+ mailinglists :) Feb 20 12:20:58 or at least scan it when something goes wrong ... Feb 20 12:20:59 rwhitby: thanks for the quick response on this ... slugosle is obviously the better alternative here Feb 20 12:21:15 rwhitby: yeah ... i should have thought about that :) Feb 20 12:21:42 BTW, why "slugosle" compared to "slugosbe" or Debian/NSLU2 ? Feb 20 12:22:46 we're using NSLU2s for embedded applications where we need to communicate a lot over serial interface ... we've been having consistent trouble with using USB<->RS232 converters when using OpenSlug Feb 20 12:23:02 interesting. ok, good reason :-) Feb 20 12:23:20 at some point we opted to give DebianSlug a shot and from that point on USB<->RS232 has worked like a charm Feb 20 12:23:52 we're still not entirely sure wether or not it may have been a 2.6.17 vs 2.6.18/19 issue ... or an endianness problem ... but my guess is towards endianness Feb 20 12:25:09 and we're using Debian Etch on an NSLU2 as our native build environment for our own software ... and since Debian/NSLU2 is LE ... slugosle is the obvious platform choice for our embedded-everything-on-flash soluition Feb 20 12:33:42 uuuh Feb 20 12:33:45 building the feed: Feb 20 12:33:46 ;;; Warning: Ignoring definition for class SIMPLE-PROGRAM-ERROR Feb 20 12:33:46 Not a condition type: SIMPLE-PROGRAM-ERROR Feb 20 12:33:46 Top level. Feb 20 12:33:46 SI> Feb 20 13:22:28 NAiL: ts101 build fails on wxbase. Feb 20 13:22:49 NAiL: let me know when you've got optware full packages build on ts101, and then I'll start it on nudi. Feb 20 13:25:34 hmmmmmm .... slugosle doesn't build because of configure stage bailing out on gettext-native ....checking for C compiler default output file name... configure: error: C compiler cannot create executables anyone else can reproduce this ? Feb 20 13:28:28 rwhitby: (Assuming you read logs). I'm waiting for the build to finish. I'll filter out packages that don't work then. Feb 20 13:43:02 * NAiL crosses finger Feb 20 13:43:03 s Feb 20 13:47:06 hmmm .. looks like after starting the build over the configure failure now doesn't occur ... Feb 20 13:57:33 NAiL: you can use nohup sh scripts/report-broken.sh & to find what packages are broken. Feb 20 14:00:21 cool, thanks Feb 20 14:15:20 03oleo * r5558 10optware/trunk/Makefile: optware: enable pure-ftpd as common cross build Feb 20 15:24:34 what's the preferred method for adding single-line startup items in slugOSbe? Is there an rc.local or something comparable? Feb 20 15:27:33 hillct: I'm not sure what the optimal method is for a debian based system but for my robot code I just created a script in /etc/init.d Feb 20 15:28:39 yeah. that works, but it seems to me there should be a standard place for user init items. I was initially stunned to find no rc.local Feb 20 15:28:59 and then you link to it from the appropriate /etc/rcn.d folder. Feb 20 15:29:31 yeah, that's more of a redhat-ish distro thing though IIRC Feb 20 15:29:59 i wonder if the LSB standard defines that.. Feb 20 15:30:08 I try to restrict use of init.d scripts to use of startup scripts for actual daemons Feb 20 15:30:34 yeah, I don't like messing with the system stuff either.. Feb 20 15:30:58 I just want to run a single command and perhaps echo a few things into /proc Feb 20 15:32:35 LSB does not include mention of rc.local Feb 20 15:57:31 yeah, I'm not sure if it even defines that sort of stuff. Feb 20 16:00:43 * hillct will wait for one of the elders for guidance Feb 20 16:01:08 NAiL: ping Feb 20 16:01:44 NAiL: any thoughts os the best/most standard place ot add single command startup items? Feb 20 16:03:00 oh, looks like I'm going to take a crack at packaging OpenPBX for the slug Feb 20 16:10:20 hillct: rc.local? Feb 20 16:10:46 ;) Feb 20 16:11:08 Really, rc.local is an ok way of starting stuff like that. Feb 20 16:18:07 I don't see it called from anywhere and there isn't an empty rc.local file out there in the distro. From where should it be called? / Is it called automatically if I just create it? Feb 20 16:19:19 there should be a symlink in /etc/rc3.d/ or something Feb 20 16:34:32 NAiL: how do you tell the current runlevel? Feb 20 16:46:34 NAiL: should the initscripts package be updated to include an empty rc.local and symlink from rc3.d? Feb 20 16:47:50 I imagine this is not something we should be doing in package setup scripts for any package that needs such an init item Feb 20 16:57:58 VoodooZ: /etc/inittab Feb 20 16:58:01 default: Feb 20 16:58:37 hillct: yes, initscripts should be updated Feb 20 16:59:01 other packages should include their own init scripts Feb 20 16:59:08 rc.local is strictly for user stuff Feb 20 16:59:17 yes Feb 20 16:59:27 well Feb 20 16:59:39 I have a package that needs to add a one line item Feb 20 16:59:59 seems a bad idea to create it's own init script for the one self contained line Feb 20 17:00:08 adjtimex -f 12345 Feb 20 17:00:30 but if you want a separate file for it, that's what I'll do Feb 20 17:21:56 NAiL: by running inittab you mean? Feb 20 17:22:13 is slugos normally running in level 3 then? Feb 20 17:22:36 VoodooZ: Look in inittab, at the default entry. Feb 20 17:22:38 NAiL: ignore the last question. Feb 20 17:22:40 oh Feb 20 17:22:42 ok Feb 20 17:22:43 sorry. Feb 20 17:22:46 just about to answer yes Feb 20 17:23:12 But how would you tell the current level (as opposed to the default one)? Feb 20 17:23:22 let's say it was changed for some reason.. Feb 20 17:23:24 no idea Feb 20 17:23:31 or Feb 20 17:23:34 run "runlevel" Feb 20 17:24:29 cool. Feb 20 17:24:42 i thought it was via telinit or something obscure. Feb 20 17:26:59 NAiL: btw, is there a reason why /boot/disk as a hardcoded sleep=6 in there? This prevents turnup's -s specified delay to work. Feb 20 17:27:36 In my case it requires > 8 secs to detect the flash disk which prevents turnup memstick from working. Feb 20 17:28:06 no idea Feb 20 17:28:20 I would recommend (I don't have write access to the repo) simply commenting out the sleep=6 line or forcing a default of 10 when none are specified. Feb 20 17:31:35 can somebody with write access comment it out then? Otherwise I'll just ask rwhitby later... Feb 20 17:33:05 Remind me a bit later and I'll do it Feb 20 17:33:17 I'm in the middle of getting optware on the TS-101 Feb 20 17:33:25 (And I think it'll work this time ;)) Feb 20 17:36:12 a buddy of mine is having problems resetting a used but never-flashed slug to factory defaults Feb 20 17:36:27 I recall seeing a wiki page on the subject but can't find it now Feb 20 17:38:57 search for resetsysconf Feb 20 17:44:45 wohoo Feb 20 17:44:54 * NAiL has a custom image for the TS101 Feb 20 17:45:06 with /opt bindmounted and a script to start optware Feb 20 17:52:30 NAiL: thanks. found the boot stuff Feb 20 17:54:18 NAiL: congrats! Feb 20 17:54:32 what's the TS101? Feb 20 17:54:40 Turbostation TS-101 Feb 20 17:54:49 QNAP's low-end nas Feb 20 17:55:10 64mb ram, ppc, 16mb flash, s-ata, gblan. Feb 20 17:55:16 fanless Feb 20 17:56:37 * VoodooZ only cares about raw speed! :) Feb 20 17:57:13 I recently reflashed my apex successfully so now I feel like a real rebel and might upgrade my RAM! hehe Feb 20 18:00:48 NAiL: the wiki for the TS-101 is really nicely laid out i find. Feb 20 18:01:51 Hi VoodooZ! I see, apex is running :-) Feb 20 18:08:41 HopsNBarley: Morning Feb 20 18:08:56 * NAiL just got a firmware image with support for optware runng Feb 20 18:09:02 s/runng/running/ Feb 20 18:09:44 howdy NAiL Feb 20 18:09:56 thanks for the kernel comments yesterday. Feb 20 18:10:33 np Feb 20 18:16:36 I'll help you in any way I can ;) Feb 20 18:18:43 Sinclair73de: yes, I decided to go ahead and flash it after only 2 tests. Much nicer than the old one too. Feb 20 18:19:11 apex rules! Feb 20 18:19:43 nicer than redboot oder than a old version of apex? Feb 20 18:21:25 both! Feb 20 18:21:58 the more recent apex version have built-in FIS driver which helps. Feb 20 19:14:53 03nail * r5559 10optware/trunk/make/mysql.mk: mysql: Fix broken link Feb 20 19:26:19 03oleo * r5560 10optware/trunk/make/pure-ftpd.mk: pure-ftpd: Build a big server with almost everything Feb 20 19:32:30 03bzhou * r5561 10optware/trunk/make/erlang.mk: erlang: ERLANG_TARGET no longer platform specific Feb 20 21:44:43 * NAiL notes that a lot of packages fail on the ts101 optware-build Feb 20 22:21:50 hi Feb 20 22:39:54 03bzhou * r5562 10optware/trunk/ (29 files in 2 dirs): re-org top optware Makefile, moved all platform specific thing into platforms/ Feb 20 23:00:19 god I hate samba Feb 20 23:00:58 is anyone here any good w/ sambas user & group handling? Feb 20 23:13:00 CoreDump|home, What is the problem. I'm not an expert but have fought with it too many times. Feb 20 23:14:11 joshin: I'm looking for a way to store user+group information for samba access seprated from UNIX user accounts. Right now a samba user mus exists as unix user etc Feb 20 23:14:16 anyone running slugosBE 4.3 who can build and test the lighttpd package for me? I found a problem with libpcre when I built for 3.10-beta. The pacakge built and installed but when I tried to start lighttpd it threw an error about libpcre after parsing the config file Feb 20 23:14:17 is that even possible? Feb 20 23:14:48 probably through PAM Feb 20 23:14:55 I'm not sure. I ran into that problem and just created the users with their shell pointing to /bin/false Feb 20 23:15:23 joshin: thought so =) Thanks! Feb 20 23:15:38 that works, but if I recall correctly, the dependency of user matching to system users, is a build-time option Feb 20 23:15:57 hillct: I'll try if it builds Feb 20 23:16:55 hillct: oh wait, you are talking about the lighttp package in the latest OE .dev tree right? Feb 20 23:17:42 yes Feb 20 23:18:24 I'm just noting that I'm running SlugOS-le not be :\ Feb 20 23:18:25 well, in the OE tree you get when you setup-slugosbe Feb 20 23:18:32 ah Feb 20 23:19:02 and I'm not using The Makefile ;) Feb 20 23:19:49 heh Feb 20 23:20:29 CoreDump|home, you mean you're building it directly with #oe? OMG, you're the man! :p Feb 20 23:20:43 /#/ Feb 20 23:20:56 joshin: yeah. Builds fine out of the box Feb 20 23:21:31 I figured it would. I've just gotten lazy in my old age. :) Feb 20 23:21:33 I take it there's an OE virtual package you build to get the image. I havn't been that daring yet Feb 20 23:21:54 bitbake slugos-image Feb 20 23:22:15 and that builds LE huh? Feb 20 23:22:25 err there's more, hold on Feb 20 23:23:00 MACHINE: nslu2le Feb 20 23:23:09 DISTRO : slugos Feb 20 23:23:13 bitbake slugos-image Feb 20 23:23:15 done =) Feb 20 23:23:30 Don't forget to do a "bitbake package-meta" to update your repository. Feb 20 23:23:45 huh? Feb 20 23:23:52 Otherwise your slug won't see all the new ipkg packages you built. Feb 20 23:24:08 CoreDump|home, after the build when you're adding stuff. Feb 20 23:24:14 joshin: ahh Feb 20 23:24:29 There should be an entry in the local.conf to automatically run package.meta Feb 20 23:24:31 bitbake package-index does the same or not? Feb 20 23:25:03 it rebuilds the local feeds "Packages" file Feb 20 23:25:11 Yeah, same thing. It probably is package-index and not package-meta. I'm not logged into my OE machine at the moment. Feb 20 23:25:34 joshin: ah =) Feb 20 23:25:58 joshin: I'm handling my own feeds w/ OE for like 2 years now Feb 20 23:26:18 <- Comes from #openzaurus Feb 20 23:26:21 Cool. Feb 20 23:26:30 oe simply rocks Feb 20 23:26:39 Same here. Back before the bitbake name change I used to contribute and update packages. Feb 20 23:26:52 joshin: hehe yeah Feb 20 23:26:54 My poor little Collie machine has been retired. Feb 20 23:26:59 with "bitkeeper" Feb 20 23:27:07 Yeah, oh the fun. Feb 20 23:27:20 mine for the time beeing as well Feb 20 23:27:36 I got busy with work around the time that Kergoth got in his car accident. Feb 20 23:29:05 * CoreDump|home resumes work on the samba "wizzard" Feb 21 00:21:31 n8 Feb 21 00:35:01 03bzhou * r5563 10optware/trunk/make/nylon.mk: nylon: 1.2 -> 1.21 Feb 21 01:01:25 <[g2]> joshin that was a couple years ago right ? Feb 21 01:02:59 * rwhitby remembers the big bitbake->monotone changeover well ... Feb 21 01:03:12 <[g2]> july right ? Feb 21 01:03:21 <[g2]> july '05 Feb 21 01:23:00 03nail * r5564 10optware/trunk/platforms/toolchain-ts101.mk: toolchain-ts101: Fixup GNU_TARGET_NAME Feb 21 01:43:43 [g2], yeah. Seems longer. :) Feb 21 02:31:19 <[g2]> joshin that's forever in open-source time :) Feb 21 02:33:46 rwhitby: DISTRO_EXTRA_.... seem to work now. thanks. Feb 21 02:34:04 cool Feb 21 02:34:28 rwhitby: I noticed the angstrom image files are nor formatted for reflash tool. How do you recommend to flash? Feb 21 02:34:55 Angstrom rootfs is too big for an NSLU2 anyway Feb 21 02:35:03 oh! Feb 21 02:35:19 so I guess one has to run it off a flash/disk then? Feb 21 02:35:35 that would explain why reflash didn't recognised it as jffs2. Feb 21 02:38:20 VoodooZ: I've only run Angstrom on an NAS100d. Feb 21 02:38:25 (16MB flash) Feb 21 02:39:43 ok. Feb 21 02:39:48 same here Feb 21 02:40:02 I tried running angstrom BE.... that did not go well Feb 21 02:40:15 alright. I'll keep playing with slugos4.3 for now. Feb 21 02:40:41 and perhaps later I'll see how to customize angstrom so I can get my precious glibc 2.5. :) Feb 21 02:41:36 what's relevant about glibc 2.5 ? (I really want to know - gentoo keeps wanting me to upgrade) Feb 21 02:41:37 rwhitby: btw, I was telling NAiL today that I found a "bug" in the initrd /boot/disk script. Feb 21 02:42:08 jacques: I just want it for NPTL as it would make threading more "clean" in my code. Feb 21 02:42:34 rwhitby: There's a hard-coded sleep=6 line in there that override the turnup specified (-s) delay. Feb 21 02:42:38 glibc 2.4 has NPTL too Feb 21 02:42:45 VoodooZ: mwester was looking at that. Feb 21 02:42:54 And with my particular setup, 6 seconds is not enough for my flash to be recognised in time. Feb 21 02:43:15 rwhitby: Ah. ok. I recommend 8+ seconds. Feb 21 02:43:44 heh - I thought you were the champion of smaller delays on boot! Feb 21 02:43:50 hehehe Feb 21 02:43:57 Yeah, that's how I found it. Feb 21 02:44:14 for testing I'm forced to use flash disk unfortunately. Feb 21 02:44:19 Just not enough flash. Feb 21 02:44:41 otherwise I wouldn't bother with turnup as it adds a good 10-15 seconds. Feb 21 02:46:04 my current setup is around 45 seconds w/ turnup. Feb 21 02:47:19 ~25 with turnup flash. Feb 21 02:47:48 03rwhitby * r699 10kernel/trunk/patches/ (2.6.19/80-kexec-arm-r3.patch 2.6.20/80-kexec-arm-r3.patch): Update ARM kexec patch for new official syscall number. **** ENDING LOGGING AT Wed Feb 21 02:59:57 2007