**** BEGIN LOGGING AT Fri Oct 30 02:59:58 2009 Oct 30 04:32:01 03bzhou * r10830 10optware/trunk/make/zile.mk: zile: 2.3.12 -> 2.3.13, again Oct 30 11:08:23 hi eno, i have some questions to some optware-packages and vt4-toolchain (perhaps more packages run for vt4-device) Oct 30 16:35:28 @eno: are you there? Oct 30 16:39:12 jomaster: yes Oct 30 16:39:36 nice Oct 30 16:41:56 at toolchain-vt4.mk - .unpacked-target: libdl will be copied to staging-dir. isn't it better to add TARGET_LIBDIR to TARGET_LDFLAGS? Oct 30 16:44:34 probably yes Oct 30 16:45:18 but usually the toolchain should be self-contained already Oct 30 16:45:32 and no need to add additional -l -L flags Oct 30 16:46:19 either way is just work around Oct 30 16:46:36 toolchain don't recognize this libs-during build. Oct 30 16:46:43 thats why this cp Oct 30 16:47:18 if i add a cp vor libpthread also apache build fine Oct 30 16:47:49 perhaps more packages work, if we copy this libs or add flags. Oct 30 16:48:22 i mean lib{dl,libpthread}*so* should be in a subdir of toolchain that it can find without -l or -L Oct 30 16:48:49 probably add flags is a better workaround Oct 30 16:48:50 but apache-build don't found libpthread Oct 30 16:49:19 it's a toolchain problem Oct 30 16:50:13 ideally it should be fixed within toolchain build Oct 30 16:50:35 but for now, I agree adding flags may be a better workaround Oct 30 16:51:08 Hey all, I'm trying to resolve a power issue with a NSLU2, anyone help? Oct 30 16:51:53 "TARGET_LDFLAGS = -L$(TARGET_LIBDIR) -Wl,-rpath-link,$(TARGET_LIBDIR)" ? Oct 30 16:52:21 "TARGET_CUSTOM_FLAGS= -pipe -I$(TARGET_INCDIR)" ? Oct 30 16:52:30 toolchain is a binary Oct 30 16:52:41 No one alive in #nslu2-general... Oct 30 16:53:11 jomaster: somebody built that toolchain binary Oct 30 16:53:25 :-D Oct 30 16:53:42 and the biggest problem of vt4 toolchain IIRC is the lack of shared libstdc++ Oct 30 16:53:45 flags ok? i don't know such thinks Oct 30 16:55:08 ah, i wondered why no libstdc++ needed :-D Oct 30 16:55:27 sorry to bother, have a good one Oct 30 16:55:39 jomaster: I'm checking Oct 30 16:55:50 looks reasonable Oct 30 16:56:25 lack of libstdc++ .so is the reason why vt4 has so many BROKEN_PACKAGES Oct 30 16:56:42 ok, if do some checks you should add this flags. Oct 30 16:57:07 why no openldap for vt4? for me it builds without an error Oct 30 16:57:07 you can go ahead add the flags Oct 30 16:57:15 ok Oct 30 16:57:46 i spent way more time demoting packages for vt4 already Oct 30 16:58:49 i don't want to be vt4 maintainer when I don't even have vt4 device Oct 30 17:08:40 03jomaster * r10831 10optware/trunk/platforms/toolchain-vt4.mk: toolchain-vt4: workaround for lib-dir (add TARGET_LDFLAGS) Oct 30 17:23:10 jomaster: i was doing a test build, and ncurses build is broken with your latest change Oct 30 17:23:39 i suspect that there's old version of ncurses in TARGET_LIBDIR Oct 30 17:24:13 there is Oct 30 17:25:36 renaming it to see if that helps Oct 30 17:26:11 the toolchain is not complete and not clean Oct 30 17:27:02 yes, that helped Oct 30 17:33:25 jomaster: apache builds finished fine Oct 30 17:33:33 with that change Oct 30 17:33:56 OTOH, you probably want to delete/rename those non-core libs Oct 30 17:34:35 such as libjpeg, libpng, libtorrent, libuuid, libz Oct 30 17:34:55 libtorrent definitely has no business in toolchain Oct 30 17:51:19 Hi I managed to compile and boot kernel 2.6.16 using the information provided here http://peter.korsgaard.com/articles/debian-nslu2.php Oct 30 17:52:30 I've tried to manually patch the latest kernel, by looking at the patch file the slugos used back then, however things seem to have changed a bit Oct 30 17:52:50 has anyone compiled 2.6.3x kernels for NSLU2? Oct 30 18:22:11 Hi Reedy Oct 30 18:23:15 Hey eno Oct 30 18:23:58 i upgraded to 9.10 last night, and could build nslu2/toolchain Oct 30 18:24:13 as long as you symlink /usr/bin/gcc to gcc-3.3 Oct 30 18:24:49 Ahhh Oct 30 18:24:56 an attempt was made to use the default gcc-4.4 as HOST_CC Oct 30 18:25:10 I did poke about with that, but it "seemed" right, to someone unfamiliar Oct 30 18:25:26 i got over the first hurdle, but failed aferwards Oct 30 18:25:31 lol Oct 30 18:27:01 Hmm, would be worth nothing that in the wiki then about gcc Oct 30 18:27:14 ok, i do Oct 30 18:27:33 right, could you test and edit the wiki please Oct 30 18:31:30 me? Oct 30 18:32:15 i'll have to do it later, gotta go out when i've eaten Oct 30 18:32:22 but i did upgrade to 9.10 also last night Oct 30 18:32:39 k, no hurry Oct 30 18:33:06 be good if we can get it updated for others though Oct 30 18:34:06 which wiki page were you following? Oct 30 18:34:24 jomaster: did you update some wiki already? Oct 30 18:34:29 @eno: also libcrypto? Oct 30 18:35:31 jomaster: right, that's part of openssl Oct 30 18:36:14 then also libssl. thats all? i'am not too familiar with the libs Oct 30 18:46:28 jomaster: yes, although i'm not too sure whether that's all Oct 30 18:47:22 libcurl too Oct 30 18:47:34 libid3tag Oct 30 18:49:27 now I know why we copied instead of adding -L Oct 30 18:50:38 better copy? Oct 30 18:51:56 why openldap is demoted? for me it works without new workaround too? Oct 30 18:54:42 jomaster: i don't remember Oct 30 18:54:52 ok Oct 30 18:55:05 you can take the approach of promoted it in platforms/packages-vt4.mk Oct 30 18:55:16 and see if it builds on builder Oct 30 18:55:33 watch http://logs.nslu2-linux.org/buildlogs/ Oct 30 18:56:04 another question to mysql5: why confirure-target don't depends on mysql5-hostbuild, but as mysql-package the build target need the hostbuild? Oct 30 18:58:02 in mysql4, later version we used gen_lex_hash and comp_err from host build Oct 30 18:58:12 not sure about mysql5 Oct 30 18:58:23 haven't touched it for a while Oct 30 18:58:41 it would be nice to upgrade to mysql5 Oct 30 18:59:35 if you're using git-svn, you can create a separate branch for experiment Oct 30 19:03:34 jomaster: the host build was added between 4.1.20 to 4.1.22 Oct 30 19:04:18 i think it is still need for mysql build Oct 30 19:05:07 mysql5.mk was probably before 4.1.22 Oct 30 19:05:21 feel free to add host build to mysql5 Oct 30 19:07:22 i will create a branch after finishing my firmware-build-system. still same weeks of work... Oct 30 19:13:14 openldap builds fine on a devel machine Oct 30 19:17:33 eno? Oct 30 19:18:12 sane-backends - 1.0.20+git20091022-1 - SANE is a universal scanner interface Oct 30 19:18:12 sane-backends - 1.0.20+git20090927-4 - Oct 30 19:18:27 Thats what I saw on optware server Oct 30 19:20:53 why there are two versions Oct 30 20:13:09 carterb: you installed an ipk from file Oct 30 20:13:20 so one is from feed, one from file Oct 30 20:13:58 i did a "ipkg list | grep "sane" Oct 30 20:13:58 the latter one (from file) does not have associated meta info Oct 30 20:14:40 this is your local ipkg's knowledge of available sane-backends packages Oct 30 20:15:06 ah, no its ok Oct 30 20:15:24 if you create an index (Package*) of your local builds Oct 30 20:15:25 now Oct 30 20:15:28 i mean Oct 30 20:15:30 sorry Oct 30 20:15:54 Can we update ejabberd to 2.1.0-rc2 Oct 30 20:15:55 then there will be two entries, both with description Oct 30 20:16:02 http://www.ejabberd.im/ Oct 30 20:16:25 we usually don't do snapshot or rc Oct 30 20:16:35 but there can be exceptions Oct 30 20:17:05 ok, I do not get the ejabberd 2.0.5 running with erlang 13 Oct 30 20:17:15 also not with erlang 12;-) Oct 30 20:17:29 sometimes the version number can be a problem too Oct 30 20:17:32 but ejabberd rc2 with rc2 works Oct 30 20:17:54 ipkg is not smart enought to tell 2.1.0-rc2 < 2.1.0 Oct 30 20:18:12 ok Oct 30 20:18:49 sorry, I am writung shit Oct 30 20:19:05 do you know when 2.1.0 is gonna be released? Oct 30 20:19:28 "but ejabberd rc2 with rc2 works" shoul be with erlang 13 Oct 30 20:19:31 close to final Oct 30 20:19:32 for your own usage, you now know how to build any version Oct 30 20:19:37 whatever that means Oct 30 20:19:46 I will take a look at the homepage Oct 30 20:20:23 "If no critical bugs are found, ejabberd 2.1.0 final may be released in one week" Oct 30 20:20:29 Should we wait Oct 30 20:20:36 yes, good idea Oct 30 20:20:42 kk Oct 30 20:20:46 thanks for testing Oct 30 20:20:49 post was from 2009-10-20 Oct 30 20:20:55 should be ready soon Oct 30 20:21:06 feel free to take over MAINTAINER Oct 30 20:21:10 I'll have a look at the page Oct 30 20:21:44 Yes, When final is released, I'll bump it and take over maintainer Oct 30 20:23:42 i often use separate git branch for different package Oct 30 20:24:09 highly recommend to use git-svn Oct 30 20:24:19 as svn client Oct 30 20:24:28 I do not understand Oct 30 20:24:33 why do you do so Oct 30 20:24:47 git is more powerful than svn Oct 30 20:24:54 whats meant by "branche" Oct 30 20:25:17 separate line of development Oct 30 20:25:18 can I do all the things you told me about svn, alo with git Oct 30 20:25:28 yes, and more Oct 30 20:25:42 e.g., you can have a sane-backends-branch Oct 30 20:25:52 ok, I think I have to read some more guidlines Oct 30 20:25:56 and an ejabberd-2.1.0 branch Oct 30 20:26:51 I'll do my best to get involved ;-) Oct 30 20:26:55 So far, in Germany we have 21.20, time to go out with my girlfriend and have some fine cocktails. Oct 30 20:26:58 great Oct 30 20:27:07 have fun Oct 30 20:27:17 thx, bye Oct 30 20:27:24 bye Oct 30 20:36:30 03bzhou * r10832 10optware/trunk/make/py-cheetah.mk: py-cheetah: 2.2.2 -> 2.4.0 Oct 30 20:39:38 @eno: libc-dev package use rsync. is this really necessary? Oct 30 20:41:14 it's convenient, we have rsync on official build machine any way Oct 30 20:41:29 if you want to replace it with cp, fine with me Oct 30 20:41:59 but libc-dev is shared with a couple of other platforms Oct 30 20:42:11 don't break other platforms Oct 30 20:42:35 perhaps tomorrow. apache and all depending packages builds fine with workaround. Oct 30 20:43:36 personally i'd leave rsync Oct 30 20:43:56 alone Oct 30 20:44:52 i don't like too much required tools at my build-system so i will remove it and do same for optware when it works. Oct 30 20:46:03 there're quite some symlinks that you need to make sure do not get copied over as another copy Oct 30 20:46:27 what linux box does not have rsync? Oct 30 20:46:48 now that I think of it, I strongly prefer to keep rsync Oct 30 20:47:11 :-D Oct 30 20:47:25 i remember reading man page of cp, and it was not obvious to do what I want Oct 30 20:47:45 it's working, let's not fix it Oct 30 20:47:47 yeah, it is great but just another depending. Oct 30 20:48:13 rsync is good tool Oct 30 20:48:54 i maybe able to simulate that with tar and untar Oct 30 20:49:03 but i did not want to bother Oct 30 20:51:42 ok, for my build-system i like to build all required tools by myself as openwrt do. this prevent build errors. Oct 30 20:52:04 but this will implemented as last step. Oct 30 20:54:30 sure, I understand Oct 30 21:01:17 e.g. if i package require automake1.9 and it is missing, only a warning is printed and build continue. Oct 30 23:43:13 good night. it's late here... Oct 31 01:14:09 03bzhou * r10833 10optware/trunk/make/py-lxml.mk: py-lxml: 2.2.2 -> 2.2.3 **** ENDING LOGGING AT Sat Oct 31 02:59:57 2009