**** BEGIN LOGGING AT Tue Jan 06 02:59:56 2009 Jan 06 03:06:08 Bartman007: any idea what might be going on? Jan 06 03:09:44 Is there a difference between the package fftw3 and the library fftw3? Are they the same? Jan 06 03:15:43 ah, this seems to be working better: make package/feeds/packages/fftw3-clean Jan 06 03:23:10 dbro: sorry, was eating dinner. Jan 06 03:24:34 oh, yeah, sorry about that mistake, I was basing it on buildroot output; I don't use scripts/feeds for my development directories, just snapshot generation. Jan 06 03:25:07 no prob. it's compiling fftw3 right now. might take a while Jan 06 03:38:08 ok, so the compiling of fftw3 finished. and I can see fftw3.h is now in the /usr/include off the staging dir. but there is not anything new in the /usr/lib - is that OK? Jan 06 03:41:27 shouldn't there be something called "libfftw3.so" ? Jan 06 03:42:13 when compiling my app, I get a different error message from before. it says: Jan 06 03:42:14 ../../../../mipsel-linux-uclibc/bin/ld: cannot find -lfftw3 Jan 06 03:42:26 my patch didn't place anything in usr/lib/, I was just working of the error message you gave. one second. Jan 06 03:42:38 righto, that makes sense. Jan 06 03:43:34 edit the fftw3 makefile to make the Build/InstallDev section look like this: http://openwrt.pastebin.com/d200f7e64 Jan 06 03:43:52 (just add the second two lines to the definition) Jan 06 03:44:22 ok, then compile fftw3, right? Jan 06 03:45:39 or can I skip all that compiling and go straight to installing? would this work: make package/feeds/packages/fftw3-install Jan 06 03:46:18 it would automatically rebuild it because you makefile has changed, yay intelligent build system Jan 06 03:46:49 it's a secret plot to get you to build a faster machine. Jan 06 03:47:27 haha Jan 06 03:47:44 ok, back to the grind... Jan 06 03:51:34 whoops. needed to change the spaces to tabs... Jan 06 03:52:12 arg, that's my fault. Jan 06 03:52:48 copying from a terminal doesn't work as it should. Jan 06 03:52:54 no biggie. might be nice to have a more helpful error message in the make process, though Jan 06 03:53:38 alright. I'm going to make some dinner while it's working. I'll be back later. and thanks very much for your help, 007 Jan 06 03:53:42 talk to the gnu make people about that :) Jan 06 04:20:18 Bartman007: I made a small change to the text you sent. the line for the copying the lib stuff had a "include" instead of a "lib" Jan 06 04:20:41 so the files now look like they're in the correct places. Jan 06 04:20:43 dbro: oops. so it did. Jan 06 04:21:09 but I'm still getting an error- I'll try to work on my program's makefile and see if I can fix it... Jan 06 04:41:06 sorry to say, it's still not working. but those error messages from before are gone. Jan 06 04:41:31 now I see a few things like this: undefined reference to `fftw_malloc' Jan 06 04:42:05 which suggest that the library is being found OK, but it doesnt contain these functions. is that the correct interpretation? Jan 06 04:42:31 and I also get one of these: collect2: ld returned 1 exit status Jan 06 05:25:16 Bartman007: It's working now. thanks! (I needed to fix some problems with my program's Makefile) Jan 06 05:28:21 \o/ Jan 06 05:29:45 yeah. phew! nice to get it going Jan 06 05:30:03 so those mods to fftw3 worked? Jan 06 05:30:12 (once you tweaked it) Jan 06 05:30:15 yes, they got the libraries in the right place Jan 06 05:30:54 and then I needed to change my makefile a bit -- for some reason the openwrt build process was more picky than the regular process to build for my desktop pc Jan 06 05:31:18 cross-compilation often can be. Jan 06 05:31:42 yes, but to get a $20 router to do all this is pretty cool Jan 06 05:32:22 agb * r13879 /packages/libs/fftw3/Makefile: [packages] libs/fftw3: install headers and libraries into staging_dir Jan 06 05:33:07 wow, the CIA knows everything! Jan 06 05:34:17 i want to know why make doesn't have an option to output a graphviz .dot file Jan 06 05:35:58 http://search.cpan.org/~agent/Makefile-GraphViz-0.18/lib/Makefile/GraphViz.pm Jan 06 05:46:06 russell_: heh, something tells me that buildroot-ng makefiles won't work with that. Jan 06 05:46:14 yeah Jan 06 05:46:50 i seriously need a clue-stick-beating when it comes to openwrt makefile hackery Jan 06 09:20:17 florian * r13880 /trunk/docs/ (Makefile adding.tex debugging.tex openwrt.tex): Make the doc slightly more complete and add notes on how to add a new target in OpenWrt, some serial console and JTAG tips and tricks Jan 06 10:19:59 mirko * r13881 /packages/lang/ (3 files in 3 dirs): fix dependencies in python EFL bindings Jan 06 11:28:03 florian * r13882 /trunk/target/linux/generic-2.6/ (config-2.6.24 patches-2.6.24/975-crypto_kconfig_hacks.patch): Prevent crypto aead from failing to build with 2.6.24 targets (#4426) Jan 06 13:40:26 MateIn4: ping Jan 06 13:40:49 johnrw: pong Jan 06 14:02:57 hi blogic... Jan 06 14:03:08 johnrw? Jan 06 14:05:00 ok... well yesterday... I was trying to work with Network Shares... Jan 06 14:05:59 and that went into a long discussion with MateIn4 about how smbpasswd does not seem to honor the adduser switch with -a Jan 06 14:06:17 yes Jan 06 14:06:32 it uses a different param i think Jan 06 14:08:10 My own personal solution was to create a ticket here where I "empowered" BusyBox to be able to adduser, addgroup... but that is plainly wrong... to add users to the /etc/passwd file. https://dev.openwrt.org/ticket/4348 Jan 06 14:08:15 nbd * r13883 /trunk/package/ead/src/ (ead-client.c ead.c ead.h): ead: allow the client to override the source ip of the server, so that it can work with route filtering properly Jan 06 14:09:23 MateIn4 pointed out that Samba should be able to make it's own list... independant of the router's "known users" Jan 06 14:09:50 ? Jan 06 14:09:58 no, because we use pam Jan 06 14:10:44 doesn't that require all samba users to have an account on the router? Jan 06 14:10:52 yes Jan 06 14:10:59 this is how it is done atm Jan 06 14:11:04 feel free to change it Jan 06 14:11:05 what is the justification for that requirement? Jan 06 14:11:07 and add a -a Jan 06 14:11:14 and use a different auth scheme Jan 06 14:11:20 and have fn coding it :) Jan 06 14:11:33 because full bvlown smbpasswd ist like 1M Jan 06 14:11:41 and our replacement is like 12k Jan 06 14:11:54 so the missing ~990k are all the account f00 Jan 06 14:11:55 I'm not asking you or anyone else to do it. I'm trying to understand the thought pattern that went into it. Jan 06 14:12:07 i never said you asked me Jan 06 14:12:23 i am just pointing out, that the reason is, that we save space this way Jan 06 14:12:35 but sure, the easiest was would be to add a new auth backend Jan 06 14:12:36 thanks, that explains a lot! Jan 06 14:12:45 there is a folder where the backends are Jan 06 14:12:47 btw. you should do a samba update, there was a security fix Jan 06 14:12:53 lol Jan 06 14:12:55 default config is not affected by the security bug, but still Jan 06 14:12:56 I know Jan 06 14:13:02 the patch is against some old samba rev Jan 06 14:13:12 we should backport the fix Jan 06 14:14:10 oh, wait. we're using samba 3.0, not 3.2 Jan 06 14:14:15 that one is not affected, iirc Jan 06 14:14:36 btw. why do we still have samba2? does anybody need it? Jan 06 14:14:52 another good question! Jan 06 14:16:39 maybe even the samba4 embedded stuff has moved forward Jan 06 14:16:47 ok well speaking as an "end user" I just gotta say... 1) Samba is very important. 2) I can't seem to find a way to make a network share writable using the provided LuCI GUI... 3) My needs are probably like most end users... where I want to mount a thumb drive and read and write to it... for anyone on the whole subnet... Jan 06 14:18:07 johnrw: I could not even find a way to make it writeable via command line Jan 06 14:18:28 I think a UI with that goal Like a pair of checkboxes... Make this Share R/W [ ] R/O [ ]? Jan 06 14:19:25 doesn't the UCI option readonly no work? Jan 06 14:19:42 +grammar Jan 06 14:19:50 Me either... and I spent the greater portion of last night with someone who had no idea how bad the breakage was... and "tried to" use the gui... Jan 06 14:20:06 the smb.conf looked perfectly well, even smaba itself reported writeable share in syslog but I simply was unnable to create files and stuff Jan 06 14:20:45 ok so an underlying samba issue Jan 06 14:20:54 It partially stems from... If Read Only is unchecked... the Luci just removes the entry... instead of placing Read Only NO Jan 06 14:21:05 johnrw: i fixed this Jan 06 14:21:11 few days ago Jan 06 14:21:33 oh yeah? Jan 06 14:21:38 ok... Jan 06 14:21:40 on 1st of january Jan 06 14:21:59 That's around when I did my last pull Jan 06 14:22:13 I might have pulled after that even... Jan 06 14:22:29 do you build from trunk or 8.09? Jan 06 14:23:19 MateIn4 suggests using tools from "out there" to actually configure Samba... since as blogic pointed out... "it's complicated" I build from trunk Jan 06 14:23:47 then at least the luci stuff should produce the right UCI file for samba Jan 06 14:25:29 to the writeable issue... older forum posts involve esotheric stuff like mount options and foo but I can't see how a mount option prevents a valid users to write to /tmp via samba Jan 06 14:26:29 but maybe it's just the gnome vfs that fools me Jan 06 14:27:05 vfs == virtual fool system? :) Jan 06 14:27:27 xMff: pff gnome... :-P Jan 06 14:27:35 likely :) (I just say: "Error details: Success") Jan 06 14:27:57 CyrusFF: pfft, you can't even set the right resolution with your KDE4 :P Jan 06 14:28:48 curse the X.org guys they do not support real multi monitor support with stupid xrandr only this totally useless "one big virtual desktop"-mode Jan 06 14:29:07 haha Jan 06 14:29:40 and I blame samba, not gnome ;) Jan 06 14:29:50 * nbd thinks both gnome and kde suck Jan 06 14:30:12 my linux machine runs xfce ;) Jan 06 14:30:28 maybe i'll try lxde soon as well Jan 06 14:30:31 looks interesting Jan 06 14:30:46 ummn I agree that both gnome and kde suck but xfce does not really suck less Jan 06 14:31:21 it does suck less, because it has much less bloat ; Jan 06 14:31:22 ;) Jan 06 14:31:42 florian * r13884 /trunk/package/kernel/modules/wireless.mk: Fix prism54 firmware indentation and malformed comments Jan 06 14:31:44 I just want sth. with one big button for xterm and alt-f2 to run programs, the rest is bloat :-> Jan 06 14:32:10 it always claims to be more lightweight than kde but I never recognized any performance advantages om xfce Jan 06 14:32:47 lars * r13885 /packages/lang/pygobject/ (3 files in 2 dirs): Jan 06 14:32:47 Replace generate-constants which runs on the host and will generate wrong Jan 06 14:32:47 constants with a target python module with the constants compiled in. Jan 06 14:36:04 CyrusFF: randr 1.3 is supposed to fix that Jan 06 14:38:19 i hope it will and hopefully it will support separate power management for monitors, its kind of annoying closing the laptop and X disables both the internal AND the external monitor with dpms Jan 06 14:38:37 because they appear to be the same "display" Jan 06 14:41:49 ok... so we agree that /etc/passwd needs to be used... to keep a list of allowed users to access network shares... my next question is how to get the username added? I had to force busybox with custom options...as discussed in https://dev.openwrt.org/ticket/4348 Jan 06 14:43:33 can someone explain the openwrt development process? is openwrt just an informal gathering for developers or is it a legal organization? couldn't find info info on web site :( Jan 06 14:44:09 it's not a legal entity yet Jan 06 14:44:19 we are working on joining SPI Jan 06 14:44:28 ok, strange that I see (C) openwrt.org Jan 06 14:47:09 1st, is it OK to discuss organizational issues in this forum? Jan 06 14:47:36 and more importantly are people interested in a discussion! Jan 06 14:48:26 list is probably a better place than forum, and yes, i do think that some people are interested in this Jan 06 14:48:57 thanks, I'll send something to the list. Jan 06 14:49:36 florian * r13886 /trunk/target/linux/brcm63xx/files/include/asm-mips/mach-bcm63xx/bcm63xx_cpu.h: [brcm63xx] correct SPI register size Jan 06 14:52:00 CyrusFF: Is there some way for me to just get the latest luci... I am building from 2.6.28 and frankly... I don't wanna touch this thing... now that I have it working right :) Jan 06 14:52:59 just doing make should get you the latest SVN version on trunk Jan 06 14:53:09 The list of LuCI packages was rather long... so is there a make package/luci-? something I can do? Jan 06 14:53:52 like make package/luci-?{clean,compile} for the samba module? Jan 06 14:54:16 make package/feeds/luci/luci-{clean,compile} V=99 Jan 06 14:54:37 xMff: thanks! Jan 06 14:54:44 (again) Jan 06 14:55:01 Someone needs to give xMff a freakin raise! Jan 06 14:55:36 I just try to explain what the others actually do Jan 06 14:56:17 yeah xMff you friggin rock. you would probably have 90% less work to do if there was decent user documentation Jan 06 14:57:12 I fear that even if we manage to have decend docs someday, they will expire pretty soon Jan 06 14:57:47 Well you are the man... with the answers... and quite frankly... without you hangin around... it would suck Jan 06 14:58:01 you need a raise. Jan 06 14:58:11 :) Jan 06 15:01:00 xMff: Jan 06 15:01:10 Will that make a new image? Jan 06 15:01:16 yes Jan 06 15:01:19 erm no Jan 06 15:01:20 great Jan 06 15:01:28 huh... Jan 06 15:01:34 ot will only rebuild all luci packages Jan 06 15:01:37 *it Jan 06 15:02:12 the magical... make image cmd? Is there one? Jan 06 15:02:20 xMff: i agree but a wiki is good against that Jan 06 15:08:45 nbd * r13887 /trunk/include/package-defaults.mk: make config.{guess,sub} writable before attempting to overwrite them (many packages have broken permissions in the tarball) Jan 06 15:08:56 danage: yes Jan 06 15:09:57 nbd * r13888 /branches/8.09/include/package-defaults.mk: merge r13887 to 8.09 Jan 06 15:28:46 nbd * r13889 /packages/net/rtorrent/Makefile: mark rtorrent as broken until somebody fixes it Jan 06 15:37:31 nbd * r13890 /trunk/package/siit/src/siit.c: fix siit compile on avr32 Jan 06 15:38:09 nbd * r13891 /branches/8.09/package/siit/src/siit.c: merge r13890 to 8.09 Jan 06 15:43:49 If somebody gets a chance, can the snapshots be updated? Jan 06 15:47:12 RoundSparrow: i think Bartman007 is working on that Jan 06 15:47:28 cool, thank you Jan 06 15:56:34 nbd * r13892 /trunk/rules.mk: add the make variable LIBGCC_S, which points to the correct libgcc ldflag (avr32 only has a static library, most targets use the dynamic one) Jan 06 15:57:02 nbd * r13893 /branches/8.09/rules.mk: merge r13892 to 8.09 Jan 06 15:59:08 nbd * r13894 /packages/net/cbtt/Makefile: use the new LIBGCC_S variable to fix cbtt on avr32 Jan 06 16:16:44 nbd * r13895 /packages/Xorg/app/eterm/Makefile: fix eterm compile on avr32 Jan 06 16:16:51 nbd * r13896 /packages/ (6 files in 6 dirs): use LIBGCC_S in more places Jan 06 16:37:01 nbd * r13897 /packages/Xorg/wm/fluxbox/Makefile: use LIBGCC_S in fluxbox as well Jan 06 16:39:49 nbd * r13898 /packages/Xorg/xorg/xserver/xorg-server/Makefile: xorg-server does not build on avr32 Jan 06 17:13:22 nbd * r13899 /packages/libs/libsndfile/Makefile: due to a compiler bug, libsndfile does not build on avr32 Jan 06 17:47:18 CyrusFF: ok... I have rebuilt Luci... and I get the same problem... if in Network Shares-> read_only is not checked... the result does not put option 'read_only' 'no' in /etc/config/samba Jan 06 17:48:24 The option itself just gets removed... but does not encourage writability by doing so... Jan 06 17:51:23 ie, it is still read only... perhaps the option could be changed to Read/Write? That way... if checked... it would insert the option 'read_only' 'no' when required... by the user. Jan 06 17:54:24 johnrw: hm, do you have luci 0.8.3 ? Jan 06 17:54:39 lemme check Jan 06 17:54:50 he said he used trunk Jan 06 17:54:57 afair Jan 06 17:57:11 yup... luci-0.8.3-1 Jan 06 17:57:34 hm it's not fixed in that version Jan 06 17:57:34 then its obvious Jan 06 17:57:37 it is Jan 06 17:58:25 indeed Jan 06 17:58:34 rm dl/luci* and make Jan 06 17:59:19 ok Jan 06 18:03:50 rm dl/luci*; make package/feeds/luci/luci-{clean,compile} V=99; make V=99 Jan 06 18:03:59 should do it Jan 06 18:04:11 yes Jan 06 18:04:21 if not the have a makefile problem Jan 06 18:04:27 *we Jan 06 18:10:11 works definitely, just tested it by myself Jan 06 18:11:54 ok... so the download was old... figures... Jan 06 18:12:45 yes we did some last minute hotfixes directly in the 0.8.3 tag and did not increase the version number Jan 06 18:13:32 so... I will be facing other bugs... after I get option 'read_only' 'no' in the smb.conf...right? Someone said earlier that it still winds up being read only Jan 06 18:14:22 yes, but that is no luci issue then, the webif just writes config files in the end Jan 06 18:14:54 i understand... but just confirming I have more work ahead :) Jan 06 18:21:10 The goal is to create/update a samba wiki page... that explains the files... and how and why it is layered like it is. If I could make a suggestion... The Samba Global's seem to come from /etc/samba/smb.conf.template ; and the /etc/init.d/samba script twiddles with some values... would it be possible to make a interface similar to the /etc/opkg.conf where a direct text edit is done for /etc/samba/smb.conf ? Jan 06 18:22:37 you mean a textbox to edit the template file? Jan 06 18:22:49 precisely Jan 06 18:25:18 For example... and edit like socket options = IPTOS_LOWDELAY TCP_NODELAY SO_SNDBUF=4096 SO_RCVBUF=4096 Jan 06 18:25:44 which seems to speed up samba according to some... Jan 06 18:26:07 would be editable... Jan 06 18:26:45 I think about a generic file editor (a.k.a. textbox with file selection) because that would cover /etc/firewall.user and some others as well Jan 06 18:27:57 Well that would be welcome... indeed... but I guess my bigger problem is to determine the reason why it stays as read only... the mounting chain... Jan 06 18:29:54 Would that be 'doable'? Jan 06 18:33:06 lars * r13900 /packages/libs/libffi/ (. Makefile): Add libffi. Jan 06 18:34:18 lars * r13901 /packages/lang/python/Makefile: Fix python ctypes module compile. Jan 06 19:30:32 has anyone tried kexec-tools on arm? seems that it's only available for x86, ps3, and mips Jan 06 19:32:36 MateIn4: I tried kexec on ARM Jan 06 19:33:03 does it work? Jan 06 19:33:07 yes Jan 06 19:33:29 http://guug.org/nit/kexec/ Jan 06 19:33:40 so there's no need to restrict to those targets in package/kexec-tools? Jan 06 19:34:32 well, kexec-tools needs some patching Jan 06 19:34:47 depending on which arm it runs Jan 06 19:35:04 I'll take a look at your patches. I'm interested in the ixp42x Jan 06 19:35:18 ok Jan 06 19:35:52 my patch is for the Nokia Internet Tablets (OMAP1 and OMAP2) Jan 06 19:36:22 thanks, which version of the kexec-tools is that against? Jan 06 19:37:18 latest on kernel.org but I don't remember right now Jan 06 19:38:16 ok, I guess that's 2.0. the kexec-tools package looks like is slightly older. Jan 06 20:11:00 nbd * r13902 /trunk/package/ead/src/ead.c: ead: fix array overrun Jan 06 20:24:45 CyrusFF: tthank you! Jan 06 20:25:04 xMff: thank you 2 Jan 06 20:25:12 for what? Jan 06 20:25:31 it does now set the read only variable... to the expected values... Jan 06 20:25:37 ah okay Jan 06 20:25:49 and are you actually able to write? Jan 06 20:25:57 now i can begin trying to make the rest of samba use those... Jan 06 20:26:36 i haven't tried... I just got it in.. and turned the setting on and off... and checked the conf file... Jan 06 20:27:10 each time... and it maintains a read only variable... in either case of true/false Jan 06 20:27:29 so with that in hand....... Jan 06 20:27:47 I can dig deeper... Jan 06 20:28:55 Frankly... i am not expecting to be able to write... as there is mounting issues that are probably to blame for the end result... Jan 06 20:34:58 xMff: nope Jan 06 20:36:24 can't write... even though it's rw in the settings; I am using the stock /tmp to test. Jan 06 20:36:41 exactly the same here Jan 06 20:40:11 whats actually the nick of john c. ? Jan 06 20:40:36 blogic Jan 06 20:41:13 ah right, thx Jan 06 21:53:49 jow * r13903 /trunk/package/base-files/ (Makefile files/etc/init.d/boot): package/base-files: fix chmod of /tmp/.uci, bump package revision Jan 06 21:56:23 jow * r13904 /branches/8.09/package/base-files/ (Makefile files/etc/init.d/boot): Merge r13903 to 8.09 Jan 06 22:20:01 nbd * r13905 /trunk/package/ead/src/ (ead-client.c ead.c ead.h): ead: add support for instance ids to prevent interference from packet reception on multiple interfaces Jan 06 22:20:12 nbd * r13906 /trunk/package/ead/src/tinysrp/Makefile.in: ead: fix spurious autoconf/automake reruns Jan 06 22:20:34 nbd * r13907 /trunk/package/ead/src/ead-client.c: ead: allow a longer timeout in the client for the prime number calculation (useful for slow devices) Jan 06 22:34:30 lars * r13908 /packages/libs/libffi/Makefile: Add libtool fixup to libffi. Jan 06 22:36:44 nbd * r13909 /trunk/ (36 files in 14 dirs): Jan 06 22:36:44 add initial 2.6.28 support for brcm47xx target Jan 06 22:36:44 Signed-off-by: Hauke Mehrtens Jan 06 22:38:58 nbd * r13910 /trunk/target/linux/brcm47xx/ (44 files in 8 dirs): nuke old 2.6.23 code for brcm47xx Jan 06 23:09:13 xMff: on trunk stream of luci - got an error saving opkg configuration (got it running on a VirtuaBox target) Jan 06 23:09:42 hi too ;) Jan 06 23:09:49 xMff? Jan 06 23:10:06 hi :) Jan 06 23:10:13 frogonwheels? Jan 06 23:10:14 * Bartman007 kicks wrt Jan 06 23:10:30 Bartman007? Jan 06 23:10:31 beat me to it Bartman007 by /that/ much Jan 06 23:11:02 xMff: /usr/lib/lua/luci/fs.lua:98: bad argument #1 to 'write' (string expected, got nil) Jan 06 23:11:03 xMff: it keeps butting into other people's conversations Jan 06 23:11:21 wrt_: bad bot. Jan 06 23:11:46 Bartman007: I just tried to imitate it :) Jan 06 23:12:05 heh, I missed that Jan 06 23:12:05 frogonwheels: okay, I'll look Jan 06 23:14:44 frogonwheels: I looked at your mail today but for me it looks like you just tarred a bunch of symlinks Jan 06 23:15:04 xMff: ?? some should have been symlinks Jan 06 23:17:54 xMff: what was the name of the tarball? Jan 06 23:18:28 lemme check Jan 06 23:19:08 found it the email - should have been 5 attachements. Jan 06 23:19:38 frogonwheels: okay, I indeed looked at the wrong places Jan 06 23:23:10 but asteriskpackage.tgz is indeed broken Jan 06 23:23:18 xMff: and sure - that asteriskpackage is just a symlink - my bad **** ENDING LOGGING AT Wed Jan 07 02:59:57 2009