**** BEGIN LOGGING AT Tue Aug 11 02:59:57 2009 Aug 11 03:06:45 xMff: I think there is some bug in spi-gpio or KS8995 module. It works first time and then not Aug 11 03:42:45 xMff: r u there? Aug 11 03:47:29 [florian]: can you svn rm feeds/packages/multimedia/uvc-streamer? #17003 only deleted the contents of the files Aug 11 04:13:19 swalker: done. Aug 11 04:13:22 agb * r17222 /packages/multimedia/uvc-streamer/: [packages] multimedia/uvc-streamer: delete leftover files and directory Aug 11 04:45:12 agb * r17223 /packages/ (3 files in 3 dirs): [packages] delete empty patch files Aug 11 07:14:09 good morning Aug 11 07:29:54 That's it. I'm fixing gspca. :) Aug 11 07:31:44 Any complaints? Aug 11 09:55:02 Bartman007: you are working on the imagebuilder? Aug 11 10:25:16 Hi. Aug 11 11:07:57 Bartman007: ping Aug 11 11:47:13 ok - next fun issue: package/feeds/packes/db I'm getting this: http://pastebin.com/d5abe8040 Aug 11 11:47:58 problem with: multiple definition of `_init' Aug 11 11:51:10 config.og: http://www.pastey.net/120800 Aug 11 12:03:00 those crti.o and crtbeginS.o are added by libtool - which I'm not really familiar with yet - I _think_ the libtool is generated by ./configure ... but don't know where the crti.o comes from Aug 11 12:04:35 hm isn't it unusual that crt* is linked into a dynamic library? Aug 11 12:05:06 # Dependencies to place before the objects being linked to create a Aug 11 12:05:08 # shared library. Aug 11 12:05:09 predep_objects="/home/michaelg/OpenWRT/build/trunk_x86/staging_dir/toolchain-x86_64_gcc-4.1.2_uClibc-0.9.30.1/usr/lib/gcc/x86_64-openwrt-linux-uclibc/4.1.2/../../../../x86_64-openwrt-linux-uclibc/lib/crti.o /home/michaelg/OpenWRT/build/trunk_x86/staging_dir/toolchain-x86_64_gcc-4.1.2_uClibc-0.9.30.1/usr/lib/gcc/x86_64-openwrt-linux-uclibc/4.1.2/crtbeginS.o" Aug 11 12:05:12 that's from libtool! Aug 11 12:05:30 xMff: I'm really unclear as to how it gets there Aug 11 12:07:52 ah the crt* stuff seems to be normal in C++ Aug 11 12:08:27 mm. ok. Aug 11 12:08:43 so wtf is the error for? Aug 11 12:09:44 where are the two definitions then? Aug 11 12:10:02 one is in crti.o - and the other?? Aug 11 12:10:21 http://bugs.gentoo.org/93739 Aug 11 12:11:22 looks like gcc itself will take care of the crt* stuff - since libtool appends them again in the commandline, this leads to duplicate symbols Aug 11 12:11:25 something like that Aug 11 12:11:32 huh. Aug 11 12:11:43 so what generates the libtool then? Aug 11 12:11:47 .. or those entries in it? Aug 11 12:12:07 try clearing predep_objects and postdep_objects Aug 11 12:12:22 make them empty strings Aug 11 12:13:29 all sources in the net indicate that newer libtool versions fix that Aug 11 12:14:16 ok. I _did_ do a google -but obviously not far enough. Aug 11 12:14:45 "libtool so crtbeginS.o multiple definitions of _init" Aug 11 12:15:01 this will bring up a couple of hits, all leading to the same conclusions Aug 11 12:15:14 yep. will put it into the make line. Aug 11 12:15:17 libtool is shit / you need to update / clear the variables :) Aug 11 14:05:59 nbd * r17224 /trunk/package/madwifi/patches/443-tx_drop_counter.patch: madwifi: fix frame dropped counter Aug 11 14:21:24 what is eth1 in this statement in the last -> /usr/sbin/dnsmasq-K-D-y-Z-b-E-slan-S/lan/-l/tmp/dhcp.leases-r/tmp/resolv.conf.auto--dhcp-range=lan,192.168.1.100,192.168.1.251,255.255.255.0,12h-2eth1 Aug 11 14:22:21 -2, --no-dhcp-interface= Aug 11 14:22:21 Do not provide DHCP or TFTP on the specified interface, but do Aug 11 14:22:21 provide DNS service. Aug 11 14:22:56 so eth1 is wan which connects to modem Aug 11 14:23:18 yes Aug 11 14:23:23 likely at least Aug 11 14:23:54 I am the same guy who has prob in getting ip address to its clients in lan Aug 11 14:24:13 in my case eth0 is lan and eth1 is wan Aug 11 14:26:27 incoming dhcp requests in logread? Aug 11 14:26:45 no Aug 11 14:27:04 then it's not a dnsmasq problem Aug 11 14:27:51 can you ping the lan if you give your client a static ip? Aug 11 14:27:51 may be in spigpio module or KS8995 switch module Aug 11 14:28:03 maybe but I can't help with that Aug 11 14:28:17 I haven't tried that, I will try that in the evening, thx anyways Aug 11 14:29:22 ok Aug 11 15:25:12 hello, any one that know about chillispot sources and tun/tap driver? Aug 11 15:25:34 tuntap should be part of the kernel Aug 11 15:28:44 <[florian]> kmod-tun more precisely Aug 11 15:39:14 hello, , <[florian]>, Yea... my issues is in chillispot, have a function to shape, leaky_bucket(), it work perfect for download trafic but not for upload Aug 11 15:40:13 -ENOCONTEXT Aug 11 15:40:54 and I want find someone that have more knowledges than me, because I guess this functions is called after tun send upload data Aug 11 15:41:45 any reason not to use iptables/imq ? Aug 11 15:42:12 sometimes I have the impression chillispots want's to be a full operating system, like emacs Aug 11 15:43:03 jejejjejejjejjejejjejejej Aug 11 15:43:17 jejejejjejee Aug 11 15:44:16 the reason is because it can be setted by radius attribute for each connections Aug 11 15:44:44 in easy way Aug 11 16:32:13 soma: yes Aug 11 16:32:18 glp: pong Aug 11 16:34:57 cool. i hope you are here later, too? i'm getting visit now, but i'll try in some hours again then. i have a patch and some comments/questions Aug 11 16:35:18 yeah, I'll be around. Aug 11 16:39:55 acinonyx * r17225 /trunk/package/kernel/modules/netfilter.mk: [package] kmod-ipt-nathelper-extra: Fix kmod-textsearch dependency (#5665) Aug 11 16:52:54 Bartman007: evening Aug 11 16:53:58 morning :) Aug 11 16:56:04 Bartman007: did the nuwiki meeting happen last week? Aug 11 16:57:09 no Aug 11 16:57:49 we need to find a better date, it's always tricky for me to attend on thursdays Aug 11 16:58:29 glp: I was sleeping unfortunately spent most of the previous few days at the hospital (not for myself, and all is well) Aug 11 17:00:53 markus * r17226 /packages/net/openvpn/files/openvpn.init: Aug 11 17:00:53 added extra commands "up" and "down" to selectively start or stop Aug 11 17:00:53 single or multiple instances of openvpn. Aug 11 17:00:53 This could be used from luci to manage running instances. Aug 11 17:01:16 xMff: can you then suggest the next date? Aug 11 17:03:39 best would be something in the weekend, but most ppl. are propably not here then (it's summer after all) Aug 11 17:04:17 Saturday perhaps Aug 11 17:35:15 is it possible to use CBI on luci, on the same page, to configure two different config subtrees? E.g., configure network and dhcp on same page? How do you create the map() to return? Aug 11 17:35:52 sorbo: hi Aug 11 17:36:02 you can't mix two configs in the same map Aug 11 17:36:18 you can - however - make two maps each pointing to a different config and return them both Aug 11 17:37:51 how do you return two maps? Aug 11 17:39:12 hi! Aug 11 17:41:13 can anyone help me with this? https://dev.openwrt.org/ticket/5671 Aug 11 17:41:36 for i, map in ipairs(maps) do Aug 11 17:41:57 ok so cbi was implemented smarty - it can expect more than one map Aug 11 17:42:15 can't seem to find a page that uses this feature, though Aug 11 17:42:49 sorbo: at the very end of your model you typically have a "return m" statement Aug 11 17:43:11 there you can simply return multiple maps like: return m1, m2, m3 Aug 11 17:43:19 each a separate Map() instance Aug 11 17:43:25 xMff: I'm at a wedding on saturday Aug 11 17:43:54 glp: Friday then? Aug 11 17:45:13 xMff - thanks it works Aug 11 17:45:53 if you want to truly miy two configs, you'd have to do a lot more work Aug 11 17:45:59 *mix Aug 11 17:47:39 wait - will it work what i've just done? It displays properly, but I haven't clicked "save" to test. Or are you saying to mix them within the same heading / section? Aug 11 17:47:44 (I'm happy to have a split page) Aug 11 17:48:15 yes, I meant if you want to control two configs within the same map, like for example mixing firewall, network and dhcp settings Aug 11 17:48:20 split page works Aug 11 17:49:20 xMff: next week is probably better Aug 11 17:49:37 glp: fine with mwe Aug 11 18:02:20 there's a bug in the cbi UI (javascript?). If you change the .enabled of a Flag, it records it properly in the backend (UCI) but the UI doesn't get updated (show fields that depend on it) when checked Aug 11 18:03:09 okay, I'll look into it Aug 11 18:03:35 actually, i don't know if it's a feature. A workaround is to use depeds("1") instad of ("whateveryochangedenabledto") Aug 11 18:03:58 auth = s:option(Flag, "auth", "Authentication required") Aug 11 18:03:58 auth.enabled = "auto" Aug 11 18:04:03 user:depends("auth", "1") Aug 11 18:04:21 this works both in backend and frontend. Though i thought it had to be "auto" in the depends, intuitively. Aug 11 18:04:58 yes I see, need to check the corresponding code, haven't looked in a while Aug 11 18:17:44 depedns("auth", "0") doesn't seem to work though (and I need it). Any tips? Aug 11 18:18:29 try depends("auth", "") Aug 11 18:18:41 flag values are actually cleared by default if they're unset Aug 11 18:20:36 works Aug 11 18:22:48 I aggree the depends for flags are bit un-intuitive since they don't reflect the actual values Aug 11 18:23:07 it's fine, as long as it works (and you know about it) Aug 11 18:23:49 by the way, is there a way to change the theme for a specific CBI page? That is, not use the default theme, but a very specific one, for a particular CBI page? Aug 11 18:24:35 one way would be a call action, and then copy whatever dispatch() does, hardcoding my own theme. but that's nasty Aug 11 18:25:35 that's a kind of layering violation, but I think I can implement a flag to set themes on a per-node basis Aug 11 18:26:32 i think that'd be useful, and certainly doesn't seem like a hack Aug 11 18:28:26 oh wait, I think we have this facility already Aug 11 18:29:12 you can assign a property called "mediaurlbase" to a node Aug 11 18:29:47 values needs to be something like "/luci-static/xyz" where xyz is the theme name Aug 11 18:29:50 yep Aug 11 18:29:55 just like in the global config Aug 11 18:30:56 it should even handle the case properly where the speicified theme is missing, it will fall back to the first installed one then Aug 11 18:31:51 worx Aug 11 18:50:09 florian * r17227 /trunk/target/linux/brcm63xx/files/ (5 files in 2 dirs): [brcm63xx] more bcm63xx definition fixes, thanks AndyI Aug 11 18:50:47 florian * r17228 /trunk/package/ar7-atm/patches/180-git_headers_include.patch: [ar7] more compile fixes for sangam-atm and 2.6.31 Aug 11 18:57:05 florian * r17229 /trunk/target/linux/brcm63xx/files/drivers/spi/bcm63xx_spi.c: [brcm63xx] fix typo, INT_MASK is writable, INT_STATUS is not Aug 11 22:02:00 is there a way to control how changes are applied using CBI? Say i want to suppress / modify the way things are applied. Or perhaps batch applies if I have more than one maps / configuration sections in a single page? Aug 11 22:04:43 not in the current implementation, it uses a simple config -> init script mapping Aug 11 22:11:51 we'll switch over to uci triggers soon, but the relevant code has needs to implemented first Aug 11 22:14:15 i think you guys did a fantastic job with luci. It'd be cool if CBIs were more flexible - i.e., that you can embed more stuff / have more control over a cbi page. Aug 11 22:15:12 For example, a status page + CBI all-in-one seems quite common. Currently you can do it as a "hack" with DummyValue, and creating fake sections Aug 11 22:15:24 you could use dummy values together with custom templates Aug 11 22:16:02 thats what I'll do soon, I want to move the status infos next to the relevant forms, like interface statistics next to the ip settings etc. Aug 11 22:16:44 also, if you have the read-only / status page, when you click "appy changes" it creates the "applying changes" div to that part of the page too Aug 11 22:16:58 you can also return a Template() and Map(), then you can design the page header as you like and the only use cbi for the normal forms Aug 11 22:17:00 e.g., if you have read-only stats at top, and read-write configuration on bottom, both sections will get the "applying changes" thing Aug 11 22:17:12 oh yes Aug 11 22:17:19 your mediabase trick allows you to do headers / footers Aug 11 22:19:50 I'd do it like this: head = Template("some/header") m = Map(...) ... foot = Template("some/footer") return head, m, foot Aug 11 22:20:24 but that only affects the area within the common header or footer, so not suitable if you want to override menus or similar Aug 11 22:20:32 heh, i'm not really used to languages that return multiple things =D Aug 11 22:20:47 yea and it probably screws up with the authentication stuff Aug 11 22:24:58 don't think it would screw with the auth Aug 11 22:25:23 ah ok - sorry, it's some "inner" header and footer Aug 11 22:25:28 yep Aug 11 22:25:28 i.e., the cbi head/foot you're talking about Aug 11 22:25:36 exactly Aug 11 23:06:21 yay. db builds (I've not done anything really since we worked out the problem. sleep and all) Aug 11 23:19:06 nbd: like this? http://pastebin.com/f470031a4 Aug 11 23:25:45 xMff: When I restarted the network I got this error iptables: Bad rule (does a matching rule exist in that chain?). Aug 11 23:26:20 acinonyx * r17230 /packages/net/kismet/Makefile: [packages] kismet: Remove deprecated configure arguments Aug 11 23:30:14 acinonyx * r17231 /packages/net/kismet/Makefile: [packages] kismet: Remove dbus dependency Aug 11 23:34:59 acinonyx * r17232 /packages/utils/usbutils/ (Makefile patches/ patches/100-ids-relative-path.patch): [packages] usbutils: Fix updating of USB IDs list Aug 11 23:37:51 acinonyx * r17233 /packages/utils/usbutils/ (Makefile patches/110-gzip-update-usbids.patch): [packages] usbutils: Create gziped list when updating IDs, bump release number Aug 11 23:55:46 Do I need to compile/install static libs for my libs to the staging_dir? Aug 11 23:56:21 not static Aug 11 23:56:42 but installing to staging dir is a good idea if other programs link against them Aug 12 00:01:00 xMff: How come? Aug 12 00:01:05 In case they want to statically link? Aug 12 00:02:02 s/statically // Aug 12 00:03:25 the staging-dir/usr/lib/ is passed in the default linker flags of every package during compilation, that way a cross-compiled application will find required cross-compiled libraries to link against Aug 12 00:04:00 technically you don't need to install your libs into the staging dir Aug 12 00:04:14 it's only required if other packages depend on them Aug 12 00:22:51 Yeah, but the libfoo.a files Aug 12 00:23:04 All the lib packages seem to install them, are they really necessary? Aug 12 00:23:39 they might be required, if some program chooses to statically link against them Aug 12 00:30:04 DURRR HUUURRR Aug 12 00:31:50 hmm Aug 12 02:23:22 hey guys, I've been told openwrt does not "support" YAFFS Aug 12 02:23:43 don't you think on NAND devices YAFFS is better than JFFS2/squashfs? Aug 12 02:32:34 I'll be back at a more appropriate time **** ENDING LOGGING AT Wed Aug 12 02:59:57 2009