**** BEGIN LOGGING AT Wed Jan 05 02:59:58 2011 Jan 05 03:03:48 that worked Jan 05 03:04:21 changing the second mutex_lock to mutex_unlock works Jan 05 03:04:31 very nice Jan 05 03:07:00 build #54 of pxcab is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/54 Jan 05 03:11:33 jow * r24906 /trunk/target/linux/generic/ (7 files in 7 dirs): [generic] fix unbalanced mutex_lock/mutex_unlock in mini_fo getxattr op. Thanks Daniel Dickinson for tracking this down Jan 05 03:15:44 build #46 of xburst is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/xburst/builds/46 Jan 05 03:17:42 build #58 of ppc40x is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ppc40x/builds/58 Jan 05 03:17:43 build #54 of ppc44x is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ppc44x/builds/54 Jan 05 10:24:20 the "Stub" version libintl and libiconv break some other packages. :( Jan 05 10:29:12 one option is let 'libiconv-full' 'libiconv-stub' install to same path. not "...-full" . then make those two package "conflict" Jan 05 11:24:29 swalker * r24907 /packages/net/l2tpv3tun/Makefile: [packages] l2tpv3tun: update to 0.2, use PKG_INSTALL and MAKE_FLAGS, cleanup whitespace Jan 05 11:32:12 swalker * r24908 /packages/utils/unrar/Makefile: [packages] unrar: update to 4.0.4, link against uclibcxx, use PKG_INSTALL Jan 05 11:36:41 xiangfu: details? Jan 05 11:37:47 xMff: hmm.. still thinking. just write down what I am think. other packages are depends on "libiconv" Jan 05 11:38:25 you said the stub versions break packages, that should not happen. Jan 05 11:38:50 esp. the intl stub is just a header file with macros Jan 05 11:39:10 mini iconv is another matter but even that should work fine Jan 05 11:39:27 xMff: yes. we setup a daily build: http://fidelio.qi-hardware.com/~xiangfu/compile-log/ Jan 05 11:39:29 it had issues with C++ until yesterday but that should be solved now Jan 05 11:39:55 xMff: all "image-full_system" after 12-14 build fail. Jan 05 11:41:14 uhm Jan 05 11:41:21 the ".config" is same with 12-14, only problem is I run "./scripts/feeds update -a && ... install -a " in daily build Jan 05 11:42:53 xMff: also we need the ...-full version. Jan 05 11:43:18 but a lot of package using static path like: "-I$(STAGING_DIR)/usr/lib/libiconv/include" :( Jan 05 11:43:33 yes, thats wrong Jan 05 11:43:59 it should be symbolic, like $(LIBICONV) which is then selected depending on user prefs Jan 05 11:44:20 btw, I don't see iconv issue in your build logs, only cross compile fuckup in htop Jan 05 11:44:20 xMff: yes. Jan 05 11:44:37 whic is also fixed since a while Jan 05 11:44:46 at least I thought it is Jan 05 11:45:05 any particular reason why you need the full iconv ? Jan 05 11:45:24 for libcharset? Jan 05 11:45:51 xMff: because we don't use openwrt in router. we use it in Ben NanoNote which have 2GB nand. and many languages Jan 05 11:46:30 xiangfu: I know, but the mini iconv as of today supports exactly the same charsets the old "full iconv" did Jan 05 11:46:33 xMff: for now we have English-Chinese dictionary, Russian terminal. ... Jan 05 11:46:51 xMff: ok. thanks for the tip. Jan 05 11:47:01 I assume you not only need full iconv but also the unstripped full iconv Jan 05 11:47:10 did you actually notice runtime issues yet? Jan 05 11:47:23 like programs aborting due to unsupported coversion Jan 05 11:47:42 xMff: I ask some day off. this is the first day back to work. I will test a little then report. Jan 05 11:47:53 okay Jan 05 11:48:35 luckily most programs use iconv and intl through glib2, so we could think about makeing it switchable there Jan 05 11:50:17 xMff: sorry for abruptness. :) Jan 05 11:53:04 xMff: I think there is one option about "sstrip" and "strip" , if the full version and stub version can switch like that will be great. Jan 05 11:53:15 yes Jan 05 11:53:25 definitely Jan 05 11:53:35 but we'll need to modify all packages using iconv or intl Jan 05 11:54:12 and make them use special macros which point to the libraries Jan 05 11:54:14 yes. that is the problem. if there is a plan maybe I can help to modify code. :) Jan 05 11:55:15 in menuconfig we have a global "native language support" option which right now basically only passes --disable-nls or --enable-nls to each configure Jan 05 11:55:34 we'd need to extend that Jan 05 11:55:45 sounds good. Jan 05 11:56:30 if you at some want real i18n for programs, including *.mo files etc. it will get real hard anyway Jan 05 11:56:36 *at some time Jan 05 11:57:13 because its patched out in many packages since openwrt lacks the infrastructure for i18n, intltool, autopoint, gettextize etc. Jan 05 11:57:41 oh Jan 05 12:02:40 I have one question about i18n, I found we "--disable-nls" in all program but the GTK program still can display i18n. Jan 05 12:02:42 I thought about the nanonote deployments breifly when ripping out i18n, but as it does not reduce functionality compared to before I just did it Jan 05 12:02:55 yeah because most programs don't recognize it anymore Jan 05 12:03:02 what is the different about "--disable-nls" and the "GTK i18n" Jan 05 12:03:11 gtk i18n happens through glib2 Jan 05 12:03:24 glib2 wraps intl and iconv Jan 05 12:03:42 glib2's gettext() is a no-op, it just returns the input string Jan 05 12:03:46 but iconv works Jan 05 12:04:48 xMff: thanks. Jan 05 12:04:48 if you say "display i18n" you mean something like can render chinese glyphs ? Jan 05 12:04:56 xMff: yes. Jan 05 12:05:11 that works in any case Jan 05 12:05:40 the i18n I'm reffereing to is message catalogs and gettext, to translate program menus etc. Jan 05 12:06:24 xMff: understand now. thanks for the answer. Jan 05 12:06:36 iconv on the other hand is a charset conversion librar ok Jan 05 12:07:00 this is used by programs that for example receive iso-8859-1 data and want to convert it to utf-8 Jan 05 12:08:01 https://dev.openwrt.org/browser/packages/libs/libiconv/src/include/charmaps.h Jan 05 12:08:11 so. for example if I --disable-nls in one program, the program mean will only have "English" Jan 05 12:08:28 yes, but it can still process input in other langauges Jan 05 12:08:47 just its menus, dialogs etc. are always fixed english Jan 05 12:11:06 in the traditional router use case it does not matter, but for desktop use it will become important Jan 05 12:11:44 yes. Jan 05 13:43:44 build #45 of mpc52xx is complete: Failure [failed shell_9] Build details are at http://tksite.gotdns.org:8010/builders/mpc52xx/builds/45 Jan 05 13:47:43 ping das-Iro Jan 05 13:48:36 on das-iro2: {standard input}:19092: Fatal error: can't write process.o: No space left on device - maybe /tmp is too small? Jan 05 13:58:11 xMff: that is fixed, there was some (150GB each slave) build debree left from previous builds Jan 05 13:58:18 ah ok Jan 05 13:58:28 sorry then Jan 05 13:59:45 no problem Jan 05 14:14:19 When OpenWRT is booting through the /etc/init.d/rcS scripts, what is a proper way to detect which storage device (partition), i.e. Flash, NAND, external USB, etc., it is booting from? Jan 05 14:16:02 The reason I asked this stupid question is I need to add a simple scripts to the /etc/init.d/asterisk and /etc/init.d/freeswitch so that if it is booting off a NAND space on my DockStar, it will use and/or look for an external USB partition to store its data. Jan 05 14:18:53 <[florian]> parsing the kernel command line from /proc/cmdline looks like it would be working Jan 05 14:19:13 <[florian]> alternatively, checking the device on which / was mounted against its major would also work Jan 05 14:52:22 build #61 of brcm47xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm47xx/builds/61 Jan 05 15:16:05 build #46 of ifxmips is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ifxmips/builds/46 Jan 05 15:18:41 build #58 of ubicom32 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/58 Jan 05 15:22:43 build #61 of at91 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/61 Jan 05 15:37:19 nunojpg * r24909 /packages/utils/watchcat/ (4 files in 2 dirs): [packages] watchcat: update to release 5 Jan 05 15:37:46 nunojpg * r24910 /packages/utils/restorefactory/ (Makefile files/uci_defaults_restorefactory): [packages] restorefactory: update to release 5 Jan 05 15:54:19 build #53 of ps3 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/53 Jan 05 16:26:11 build #55 of pxcab is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/55 Jan 05 19:31:37 juhosg * r24911 /trunk/package/mtd/ (Makefile src/Makefile src/mtd.c): package/mtd: make fixtrx available on ar71xx as well Jan 05 21:28:44 ping blogic Jan 05 21:31:46 swalker * r24912 /packages/utils/i2c-tools/ (Makefile patches/100-linux_include.patch): [packages] i2c-tools: update to 3.0.3, refresh patches Jan 05 21:33:37 swalker * r24913 /packages/net/p910nd/ (Makefile patches/): [packages] p910nd: update to 0.94, drop obsolete patch, use PKG_INSTALL Jan 05 21:45:45 hm, didn't notice that .37 will feature kernel mode pptp Jan 05 21:50:51 xMff: know anything about pjsip? Jan 05 21:51:02 no Jan 05 21:52:43 cshore: what do you want to know? Jan 05 21:52:51 xMff: trying to find out how stable it is. Google doesn't have much to say, searching for stability or on another search for bugs, but it doesn't have that much to say at all really Jan 05 21:52:58 larsc: you too :) Jan 05 21:56:17 hm, i think it is quite stable, but i had some problems with some sip providers. but could have been me doing something wrong, when using the api Jan 05 22:00:11 <_trine> I have compiled 2.6.37 into my wndr3700 and it works fine Jan 05 22:01:26 <_trine> Linux OpenWrt 2.6.37 #1 Wed Jan 5 18:53:04 GMT 2011 mips GNU/Linux Jan 05 22:03:51 xMff: if you see the people could you let him know I'm not the FreeSWITCH maintainer for at least nine months due to a non-competition agreement with my new employer Jan 05 22:04:18 okay Jan 05 22:04:24 thepeople: ^ Jan 05 22:04:26 :P Jan 05 22:04:34 :P Jan 05 22:04:40 <_trine> :( Jan 05 22:05:00 job >> hobby Jan 05 22:05:18 xMff; unfortunately, yes Jan 05 22:05:24 <_trine> he will have to be executed now Jan 05 22:05:33 <_trine> at dawn Jan 05 22:05:36 but at least I can still work on openwrt Jan 05 22:05:50 luckily I work in a different field so there's no conflict with openwrt Jan 05 22:06:20 <_trine> does this mean NICo wins the job again Jan 05 22:06:36 well Jan 05 22:06:45 first he must reappear Jan 05 22:07:01 he might be overly busy....I think that's why he let me do it before he disappeared Jan 05 22:07:13 going from my logs, I would guess thepeople is currently on vacation or something ;) Jan 05 22:07:18 <_trine> {Nico}, attention (call of duty) Jan 05 22:08:15 <_trine> thepeople, according to my sources is not doing anything with webif either due to other pressures Jan 05 22:08:38 afaik fofware jumped in to do maintainance Jan 05 22:08:45 <_trine> yep Jan 05 22:08:50 maintenance ? Jan 05 22:08:54 can never spell it right Jan 05 22:09:12 <_trine> its right Jan 05 22:09:36 <_trine> maintenance is correct Jan 05 22:09:50 k Jan 05 22:11:14 hrm, atheros bought by qualcomm Jan 05 22:12:26 yes Jan 05 22:17:11 qualcomm is a member of the linux foundation, so it might not be that bad Jan 05 22:19:13 yeah, let's hope its no orcal-sun style takeover Jan 05 22:19:22 *oracle-sun Jan 06 01:30:15 build #62 of atheros is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/atheros/builds/62 Jan 06 01:36:02 lars * r24914 /trunk/target/linux/xburst/ (33 files in 6 dirs): [xburst] Add 2.6.37 support Jan 06 01:37:08 lars * r24915 /trunk/target/linux/xburst/ (10 files in 6 dirs): [xburst] Drop support for older kernel versions Jan 06 01:49:25 xMff, kaloz, nbd: now that mini_fo+ext4 works, if desired by all we can disable the ext2 and ext3 modules (though I'm a little confused as to why that dependency which resulted in that was originally created in the first place, so not convinced it's the right thing to do). Jan 06 01:50:23 ext4 can take over 2 & 3 just fine Jan 06 01:50:42 the only reason they got added back was due to the issues with ext4 Jan 06 01:50:57 xMff: ok Jan 06 02:01:57 fyi: wiki.openwrt.org & luci.subsignal.org will be down for 5-30 mins, sorry for inconvenience Jan 06 02:07:03 build #57 of orion is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/57 **** ENDING LOGGING AT Thu Jan 06 02:59:58 2011