**** BEGIN LOGGING AT Sat Feb 25 10:59:57 2006 Feb 25 12:37:17 hello. I have a short question: Is there a kernel-sources package available for DebianSlug? I didn't find one. I need to compile a special kernel module that comes not as ipkg package Feb 25 12:39:30 bwalle: debianslug from OE, or the debian kernel built in debian? Feb 25 12:39:42 from OE Feb 25 12:40:52 bwalle: we have no kernel or module debian packages coming out of OE. Your two options are: Feb 25 12:41:19 1) use the debian kernel and it's source package (note that the debian kernel is not fully functional) Feb 25 12:41:56 2) build the kernel module in OE - see the zd1211 (or something like that) as an example, or ask dyoung how to do it Feb 25 12:47:12 2) looks like a very good idea :) Feb 25 13:05:03 03repvik * r267 10/releases/slugos-3.6-beta/openembedded/packages/busybox/ (4 files in 2 dirs): Busybox: Add sysctl applet Feb 25 13:06:07 NAiL: make sure you add it to monotone head too Feb 25 13:06:25 That's a "backport" from mtn Feb 25 13:07:27 gotcha. thx. Feb 25 13:08:24 I've just pushed an updated version to mtn now (forgot to sync earlier) Feb 25 14:02:03 03repvik * r268 10/releases/slugos-3.6-beta/openembedded/packages/busybox/busybox-1.01/uclibc_posix.patch: Add forgotten patch Feb 25 14:44:19 03repvik * r269 10/releases/slugos-3.6-beta/openembedded/packages/busybox/ (busybox-1.01/slugos/sysctl.conf busybox_1.01.bb): Add hotplug fix Feb 25 15:00:25 hi Feb 25 15:01:07 the eggdrop installation has a little error Feb 25 15:01:27 it says the eggdrop.conf will be in /opt/etc/ Feb 25 15:01:39 but it is in /opt/eggdrop Feb 25 15:01:41 ?!? Feb 25 15:01:58 thats under unslung Feb 25 15:03:04 6.5 alpha Feb 25 15:03:23 but under unslung 5.5 beta its in /opt/etc as it should be Feb 25 15:03:25 hmm Feb 25 15:03:41 the ipkg packet versions are the same Feb 25 15:04:21 oh Feb 25 15:04:23 no wait Feb 25 15:04:42 under unslung 5.5 there is an eggdrop.conf in the eggdrop folder too Feb 25 15:04:50 dunno which was first :-( Feb 25 15:05:04 and where it really should be Feb 25 15:07:45 oh and the old error "./eggdrop: error while loading shared libraries: libtcl.so: cannot open shared object file: No such file or directory" is there again Feb 25 15:08:22 again the problem Feb 25 15:08:27 that after unsling Feb 25 15:08:39 the path variable is missing the opt folders Feb 25 15:09:20 that´s something that should be changed on the firmware image scource code Feb 25 15:09:43 that a /etc/profile is created after unsling with the correct path variable Feb 25 16:00:06 Lordeiffel: The path cannot be so easily changed upon unsling; certain system functions depend upon the system finding the Linksys proprietary version of busybox in /bin; this has broken when the path was updated in /etc/profile. It is recommended that you install and use ssh (ipkg install openssh) instead, as it ensures your PATH is set correctly without breaking the system scripts. Feb 25 16:04:05 ah ok Feb 25 16:08:07 Anyone with admin access to the monotone repository around today? Feb 25 16:09:20 mwester: what's the problem? Feb 25 16:10:33 Can't update monotone - I get a disconnect; maybe my key got lost (it's been a while since I checked anything in)? Feb 25 16:10:47 enter passphrase for key ID [mwester@openembedded.org]: Feb 25 16:10:49 monotone: read from fd 6 (peer monotone.nslu2-linux.org) failed, disconnecting Feb 25 16:12:15 rwhitby: ping Feb 25 17:01:14 03repvik * r270 10/releases/slugos-3.6-beta/openembedded/packages/zd1211/ (4 files in 2 dirs): zd1211: Update to latest svn rev. Feb 25 17:01:59 03repvik * r271 10/releases/OpenSlug-2.7-beta/openembedded/packages/zd1211/ (10 files in 3 dirs): zd1211: Update to latest svn rev. Feb 25 17:02:59 03repvik * r272 10/releases/OpenSlug-2.7-beta/conf/distro/openslug-bbfiles.conf: Update zd1211 feed version to -r59 Feb 25 17:05:23 03repvik * r273 10/releases/OpenSlug-2.7-beta/openembedded/packages/zd1211/ (4 files in 3 dirs): zd1211: Clean out old cruft Feb 25 17:40:37 macsat : ping Feb 25 21:03:56 Lordeiffel_: eggdrop should be compiled to search /opt/lib - there is *never* a need to include /opt/lib in the path or to change LD_LIBRARY_PATH - it should always be done at compile time in Optware. Feb 25 21:04:16 mwester: did your monotone problem resolve itself? Feb 25 21:04:23 oleo: pong Feb 25 21:04:59 rwhitby: no - I remain adrift, unable to dock with the mothership! :D Feb 25 21:05:27 ok, let me tail the log file and I'll get you to try again ... Feb 25 21:05:52 ok, try now Feb 25 21:07:34 mwester: what's the last octect of your IP address? Feb 25 21:07:54 Should be .110 Feb 25 21:08:33 have you tried again yet? Feb 25 21:09:06 Er, not just yet; slow 'puter today... Feb 25 21:10:37 monotone: failed to process '4' packet Feb 25 21:10:54 what version of monotone are you using? Feb 25 21:11:11 24 Feb 25 21:11:33 Hmm. Server is running 0.25, but that's supposed to support 0.24 clients too. Feb 25 21:11:52 I built 25, but havn't installed it yet for fear of compounding the problem... Feb 25 21:11:54 can you update to 0.25 ? Feb 25 21:12:01 I'll try right now. Feb 25 21:14:04 Same problem :( Feb 25 21:14:50 are you able to easily save your outstanding changes, and then reapply them to a new copy of the database? Feb 25 21:16:39 I kept notes; I think I can do so; there's really no intermediate versions I need. I'll blow away all, then use make update to fetch a new DB -- since the certificates are no longer in the DB, that should be all, right? Feb 25 21:17:05 yup Feb 25 21:17:45 mwester: copy the file tree and the diff -r to get a patch of the changes after you co a new copy Feb 25 21:35:06 :( No luck, new database fails with same error: monotone: read from fd 6 (peer monotone.nslu2-linux.org) failed, disconnecting Feb 25 21:36:10 same error here too Feb 25 21:36:55 * NAiL tries Feb 25 21:37:29 well "there" rather than "here, cause the server is probably closer to you than it is to me ;-) Feb 25 21:37:46 Uh-oh -- I BROKE IT! :o Feb 25 21:38:10 I've seen changes made today, though Feb 25 21:38:31 It's merged a few times when I've "updated" over the past 18 hours or so. Feb 25 21:39:34 that's expected Feb 25 21:40:58 hmm Feb 25 21:41:01 works fine here Feb 25 21:41:11 mwester: nm - your key is not in there. Feb 25 21:41:21 that must be the problem Feb 25 21:42:00 Uh-oh. At least it's not a data corruption problem, that's good. Feb 25 21:43:01 can you try now? Feb 25 21:44:20 Well, how about that! "monotone: successful exchange with monotone.nslu2-linux.org" - Thank you! Feb 25 21:44:50 sorry to put you through the hassle when it was something I should have checked first Feb 25 21:45:24 No problem; I wish there was a way I could have checked it myself! Feb 25 21:45:53 monotone list keys Feb 25 21:46:05 (on a fresh download of the db) Feb 25 21:46:39 mwester: so are there any outstanding Unslung 6.x things? Or is everything in the perfect state for a release? Feb 25 21:46:55 I see... yep, that gets the list - cool! Feb 25 21:47:45 Yes, I'm just cleaning up some outstanding Unslung 6.x things. A new defconfig for the kernel, and some rootfs and image related things. Feb 25 21:49:19 excellent. will you be updating the README and NOTES file (including the ChangeLog at the end) ? Feb 25 21:49:46 README yes, I guess I'll add the NOTES and ChangeLog as well, huh? :) Feb 25 21:50:06 please Feb 25 21:50:43 There's also a number of Wiki articles that will have to be updated. I'll be busy... Feb 25 21:51:18 well, we have 20 alpha testers who should be helping update the wiki too ... Feb 25 21:52:00 Er, should I bump it to 6.6-alpha (if so, where?), or just check things under 6.5-alpha for now? Feb 25 21:52:39 if you make changes, then bump to 6.6-alpha (numbers are cheap) Feb 25 21:52:47 Re: wiki -- my cynical side is getting the best of me this week; I suspect the wiki won't see much update unless one of us does it... Feb 25 21:53:01 in conf/distro/unslung.conf Feb 25 21:53:19 we should get the documentation sub-project onto it. Feb 25 21:53:21 Ok, 6.6-alpha coming up later this afternoon -- er, morning where you're at? Feb 25 21:54:09 8:30am here Feb 25 21:54:31 just say "in N hours" :-) Feb 25 21:55:16 :D Ah, but I've learned that "later this afternoon" gives me MUCH more lattitude than "N hours" at work! :D Feb 25 21:55:37 How about in "approx. 2 - 6 hours?" then?? :D Feb 25 21:56:09 well, I can't cut your pay any lower than it is ;-) Feb 25 21:59:03 oleo: pong Feb 25 22:00:04 hi. I have some NEWS regarding uClibc++ on wl500g Feb 25 22:02:04 And i want to talk to prime achitect if it is OK to replace nonexistent libstdc++ with uClibc++ ? Feb 25 22:05:13 Now I've rebuilded all packages with libuclibc++ instead of libstdc++ and it seems that this is a way to go. Toolchain is still hndtools-mipsel-uclibc-3.2.3-full but with g++-uc wrapper. Feb 25 22:06:25 ok, so if there was not a libstdc++ in the repo before, then all you have done is added uClibc++ and enabled more new packages right? Feb 25 22:11:46 At the moment no new promotions. But tried groff and now man works! Even got mysql running. I am not shure which packages run with libstdc++ on wl500g. I've tried some without success long time ago. I've saw that some people installed libstdc++ from openwrt and that some packages can run. No definitive answer, though. Feb 25 22:15:14 well, if it gets new packages working, and has no effect on existing packages, then I say go for it. Feb 25 22:17:35 Yes. I think that this is a way to go. But it requires to install uClibc++ in toolchain with command make libuclibc++-stage Feb 25 22:21:03 ok, so that will just be done in the staging dependencies for any tool that requires uClibc++, or should it be done once at the start as part of make toolchain ? Feb 25 22:22:42 unslung/Makefile must be adjusted to use HOST_CXX=g++-uc . I think that make toolchain is the best approach. Feb 25 22:26:11 Maybe i can somehow figure out how to set variables to use wrapper when libuclibc++ is depend in package. This will also introduce more ifeq OPTWARE_TARGET into 10 packages at most. Feb 25 22:26:50 well, you can just define HOST_CXX differently for wl500g from other targets Feb 25 22:27:12 and if it's in toolchain, then you don't need to depend on it in a package (you can assume it's there) Feb 25 22:29:03 Right. There is HOST_CC at the moment in the Makefile. No HOST_CXX. Are this variables somehow standard for make? Feb 25 22:29:16 no Feb 25 22:29:51 are you sure it's HOST_CXX you want to change? That it used to build things to run on the build host, not on the target. Feb 25 22:30:44 (i.e. are you sure you don't want to change TARGET_CXX instead?) Feb 25 22:30:53 At the moment I have simpler solution with TARGET_CXX=$(TARGET_CROSS)g++$(TARGET_CXX_WRAPPER) and define TARGET_CXX_WRAPPER=-uc for wl500g only. Feb 25 22:31:50 just put an ifeq OPTWARE_TARGET around the definition of TARGET_CXX Feb 25 22:32:14 or move it into the existing ones Feb 25 22:32:50 actually, it's best to move it into the existing ifeq sections, so it can have a separate value for each target if necessary. Feb 25 22:32:52 OK. Not my style but it conforms to others. I can do this. Feb 25 22:33:25 nod. thanks for being accommodating. Feb 25 22:34:36 will you be going through wl500g_broken_packages to see which ones now build as a result of this change? Feb 25 22:36:25 I've already done this from scratch. To verify if it works. And I see only two python packages to come out broken. Feb 25 22:36:45 so *all* the other broken ones now work? Feb 25 22:38:13 NO no. No promotions. Here is additional list of broken packages I've found: py-rdiff-backup py-scgi py-turbogears py-setuptools py-testgears py-sqlobject flac libftdi tethereal Feb 25 22:39:06 So some of those don't fail previously, but now do fail due to uClibc++ ? Feb 25 22:42:08 You can say that yes. But this was wild compile without study why was broken. I mus say that I did not tested many for real. libUclibc compile is also at the moment with dissabled EH and RTTI. Feb 25 22:44:22 Do you know if python is actually running on wl500g with libstdc++ ? Feb 25 22:44:45 don't know. Feb 25 22:45:03 I don't use wl500g optware packages very much at all myself. Feb 25 22:46:33 FYI, on nslu2, i just did a # objdump -p /opt/bin/python | grep NEEDED Feb 25 22:46:33 NEEDED libpython2.4.so.1.0 Feb 25 22:46:33 NEEDED libpthread.so.0 Feb 25 22:46:33 NEEDED libdl.so.2 Feb 25 22:46:33 NEEDED libutil.so.1 Feb 25 22:46:33 NEEDED libstdc++.so.5 Feb 25 22:46:52 NEEDED libm.so.6 Feb 25 22:46:52 NEEDED libgcc_s.so.1 Feb 25 22:46:52 NEEDED libc.so.6 Feb 25 22:47:10 maybe you can do the same on wl500g Feb 25 22:48:04 OK. I will spent a snowy sunday and will see how this can be commited. Feb 25 22:51:19 ok, thx Feb 25 22:53:14 eno : with this ruby thing when cross compiling. Also ruby on host is needed. Can you include some notice on host ruby depends. Feb 25 22:58:54 ok Feb 25 22:59:25 or maybe build a host-ruby before, just like python Feb 26 00:12:54 eno: ping Feb 26 00:47:00 oleo, pong Feb 26 00:50:32 eno: Do you know if python works on wl500g target? Now I've tried with libstdc++ without success. Feb 26 00:51:08 i don't know since i don't have a wiley Feb 26 00:52:08 file python2.4 shows python2.4: ELF 32-bit LSB executable, no machine, version 1 (SYSV), for GNU/Linux 2.2.0, dynamically linked (uses shared libs), stripped Feb 26 00:52:38 no machine Feb 26 00:53:05 what is reported on nslu2 with file? Feb 26 00:53:28 on slug it's: Feb 26 00:53:38 /opt/bin/python2.4: ELF 32-bit MSB executable, ARM, version 1 (ARM), for GNU/Linux 2.4.3, dynamically linked (uses shared libs), stripped Feb 26 00:54:18 maybe it never worked? need to ask whoever promoted it Feb 26 00:55:39 didn't look into patches, but: "Is there any ARM dependent issues?" Feb 26 00:56:13 eno: for lots of wl500g packages, just compiling was enough for promotion - it may not have been tested Feb 26 00:58:20 checking... Feb 26 00:59:28 I can say: py* py-* is broken on wl500g from day one. Feb 26 00:59:43 not surprising Feb 26 01:00:54 could not find anything arm specific right now Feb 26 01:01:42 we probably should have some mechanism like the alpha testing Feb 26 01:02:19 well, we put the packages out there, and if someone complains that it doesn't work then it is either fixed or demoted Feb 26 01:02:40 (since there is no active wl500g package manager - anyone want the job?) Feb 26 01:02:55 so we should demote py* for wiley now Feb 26 01:03:24 i thought jf is wl500g pkg manager Feb 26 01:03:56 he is, but has no time to work on it Feb 26 01:04:16 eno: yes, if it doesn't work, then add it to WL500G_BROKEN_PACKAGES Feb 26 01:04:17 i can try to get python to work on wl500g if you give me access to one Feb 26 01:04:46 remote access is fine, cannot promise on the time Feb 26 01:04:58 I wanted to know this as there is a lot of py*ckages and want to promote uClibc++ Feb 26 01:06:05 eno: my only wl500g is my internet gateway, and I cannot give you access to that. Feb 26 01:06:14 understand Feb 26 01:06:33 ka6sox might have a dev wl500g available Feb 26 01:06:39 not sure if he is using it in production Feb 26 01:06:47 eno: I can give you access on my wl500g. Feb 26 01:07:44 thx oleo, right now someone need to mark python* broken Feb 26 01:08:26 anyone with cvs write access can do that Feb 26 01:09:16 I say. It is not broken if it compiles. I can live with that. Feb 26 01:10:07 no, if it doesn't run then it is broken. Feb 26 01:10:40 the thing is that when new packages are added to optware, they are not tested on every target. If it compiles, and has not previously been marked as broken, then it is promoted by default. Feb 26 01:11:25 Yes but if you follow wl500g forum then you can see in which packages is interest. And then you can put effor on it. Feb 26 01:11:26 03bzhou * 10unslung/Makefile: mark all py* packages broken for wl500g Feb 26 01:12:14 oleo: true - that's the job of the wl500g package manager (or others who are interested) Feb 26 05:24:07 i just successfully got into redboot, on a slug that had been unslung last august. i had forgotten the ip i had set it to, and now want to set it to something new. i need basic network help to understand what the ip should be. Feb 26 05:35:38 i think i need to set it back to 192.168.1.77, and then set up a little lan off my ubuntu machine. then i could hit the admin pages. Feb 26 07:07:33 03bzhou * 10unslung/make/py-sqlalchemy.mk: adding py-sqlalchemy **** ENDING LOGGING AT Sun Feb 26 10:59:56 2006