**** BEGIN LOGGING AT Sun May 13 02:59:59 2007 May 13 03:46:07 do i need to do something else besides creating a dir with a Makefile and a Config.in inside the toolchain dir, to see the thing later in menuconfig? May 13 03:49:24 <[mbm]> what are you trying to do? May 13 03:53:18 [mbm]: maybe you read what i wrote about 4h ago, if not: i thought about the idea to add extra deps like ruby or sdcc to the toolchain, if the dependencie check can't find them it'll start to build its own tools. imho it would make it easier to reproduce bugs because the buildsystem gets a little bit more independent from the host and also it would help people to build it if they can't install the other tools May 13 03:54:55 <[mbm]> Ugh, before long we'll be compiling a linux distro May 13 03:54:59 <[mbm]> and I don't mean openwrt May 13 03:55:52 <[mbm]> only the menus for packages are auto generated, for everything else you have to follow the chain of Config.in includes May 13 03:56:25 i know that it isn't the aim of openwrt, just thought about it as kind of fallback if these things are not on the system where you are going to compile it May 13 03:58:32 <[mbm]> the problem will go away as we start to have binary releases of kamikaze, reducing the need for manually compiling May 13 04:14:12 [mbm]: is there something i need to put into configfile to get it build? got it into menuconfig by putting this line source "toolchain/ruby/Config.in" into toolchain/Config.in thanks for the tip with Config.in :) May 13 04:14:57 <[mbm]> now follow the makefiles May 13 04:15:20 ok :) May 13 04:17:29 in my mind somehow the phrase "follow the white rabbit" arise :D May 13 04:39:54 is the ":" a separator in line 36-38 in toolchain/Makefile ? May 13 04:48:33 <[mbm]> those are makefile targets May 13 04:49:00 <[mbm]> normally you'd see them written like this: May 13 04:49:04 <[mbm]> foo: foo.c May 13 04:49:10 <[mbm]> gcc foo.c -o foo May 13 04:49:38 <[mbm]> : May 13 04:49:46 <[mbm]> the command is optional May 13 04:50:10 <[mbm]> all we're doing is enforcing an order May 13 04:51:09 ok, so this binutils-prepare: uClibc-prepare means it'll prepare uclibc before it prepares binutils? May 13 04:51:19 * crazy_imp is a little bit confused May 13 04:51:44 <[mbm]> right May 13 04:52:11 <[mbm]> if you scroll down in the makefile you'll see targets like %-prepare May 13 04:52:16 <[mbm]> % is a wildcard May 13 04:52:29 i know :) May 13 06:14:21 everything in trunk/tools will be build, no way to deselect something in there, right? May 13 06:17:02 <[mbm]> you should have been paying more attention to the toolchain makefile May 13 06:18:36 ? can i get another little hint ? :) May 13 06:19:34 ah, guess i found it May 13 06:20:28 the trick with TARGETS-y / TARGETS_$(CONFIG_bla) if it's "y" it gets added May 13 06:21:15 <[mbm]> that trick appears all over the place in the linux kernel May 13 06:22:28 normally i don't read the kernelsources May 13 06:33:58 [mbm]: is it possible to get it working with something like binutils where in .config is only a CONFIG_BINUTILS_VERSION_2_16_1=y - add it x-times for each version or doing something to chop of the end of the string (both would be ugly imho). May 13 06:35:03 <[mbm]> huh? that made no sense May 13 06:41:07 for example binutils, you can only select the version, but it will be build by default, so it's "static" in TARGETS-y. but if i would have something with multiple versions i can select _and_ also the choice to build it or not, there will be by 2 versions i can choose, BLA_1 and BLA_2 in .config May 13 06:42:36 so to get the TARGETS_$(CONFIG_bla)-trick working i would need to add for each version a "TARGETS_$(CONFIG_BLA_x) += bla" line in the Makefile, that would be one way i think May 13 06:43:31 or one line with some code inside to get rid of the version-number May 13 06:43:51 <[mbm]> that's all done through the Config.in entries, you have a on/off toggle for the item and then you have a version select depend on the toggle May 13 06:44:33 <[mbm]> so you end up with a CONFIG_FOO=y CONFIG_FOO_VERSION=x.x.x May 13 06:44:47 ah, thats good :) May 13 06:46:48 <[mbm]> ... May 13 06:47:14 <[mbm]> we already had the conversation that I'm not going to add an entire linux distro just to build a linux distro May 13 06:47:20 [mbm]: thanks again for your help :) May 13 06:48:19 [mbm]: i know, i will not commit that stuff, but it's a good way to understand how it's working :) May 13 06:49:36 [mbm]: happy now? May 13 06:54:45 <[mbm]> manic depressive and you think a few lines on irc is going to change that? May 13 06:58:27 [mbm]: hmm, guess to change that is chillywilly's task :P May 13 09:03:59 florian * r7198 /packages/libs/glib2/ (Makefile patches/ patches/001-no_glib_marshal.patch): Fix glib2 compilation errors when no glib2-dev package is installed on the host system (#1659) May 13 09:06:47 [florian]: i nearly started to make a patch too :D May 13 09:39:21 mbm * r7199 /trunk/package/base-files/files/sbin/mount_root: clean up the mtd unlocking May 13 09:40:40 <[florian]> crazy_imp: thanks ;) May 13 10:17:56 florian * r7200 /trunk/package/base-files/files/ (etc/preinit sbin/mount_root): Fix set_state preinit stuff May 13 11:05:52 [florian]: your patch works fine, removed glib2 from my system and it build :) sorry that i overlooked this as i wrote the Makefile :| May 13 11:32:25 <[florian]> crazy_imp: no problem, I did that mistake too when I commited glib2 May 13 11:34:26 [florian]: but the person who reported the ticket was wrong too, it's not a target thing :D May 13 11:38:22 ryd * r7201 /packages/net/sshfs/ (. Makefile): add sshfs May 13 12:38:50 will a pack6age also rebuild if i put something into package/bla/patches ?i know it'll be rebuild if i modify the Makefile, but does it also watches the stuff inside the patches dir? May 13 12:39:08 s/pack6age/package/ May 13 12:43:22 florian * r7202 /packages/admin/gkrellmd/ (6 files in 3 dirs): Add gkrellmd (#1453) May 13 12:47:48 florian * r7203 /packages/admin/gkrellmd/ (Makefile files/gkrellmd.init): Standardize makefile and init script May 13 12:47:57 <[florian]> crazy_imp: it should yes May 13 12:48:09 [florian]: ok May 13 13:31:06 kaloz * r7204 /trunk/target/linux/ixp4xx-2.6/image/ (Makefile apex/patches/100-slugos-nslu2-armeb_config.patch): fix the nslu2 image for the new layout May 13 15:54:22 nbd * r7205 /tags/kamikaze_pre1/: branch kamikaze_pre1 May 13 15:59:33 nbd * r7206 /tags/kamikaze_pre1/target/linux/ (11 files in 11 dirs): remove unsupported targets May 13 16:01:32 nbd * r7207 /tags/kamikaze_pre1/toolchain/gcc/ (Config.in Config.version patches/4.2/): remove unsupported compilers May 13 16:03:30 nbd * r7208 /tags/kamikaze_pre1/package/ (ar7-atm/ base-files/files/etc/banner): remove ar7-atm May 13 16:04:41 nbd * r7209 /tags/kamikaze_pre1/package/base-files/files/etc/ipkg.conf: update ipkg.conf for new location May 13 16:07:29 nbd * r7210 /tags/kamikaze_pre1/package/base-files/files/etc/openwrt_version: add openwrt_version May 13 16:24:11 nbd * r7211 /trunk/package/madwifi/ (31 files in 2 dirs): huge madwifi update - use a madwifi-ng-refcount snapshot as base, includes lots of bugfixes and performance enhancements May 13 16:24:15 :) May 13 16:25:21 cool May 13 16:26:48 please test it, this was a lot of work and i don't have any long term testing results yet May 13 16:28:58 I'm building right now May 13 16:51:30 * thepeople_mobile builds May 13 17:26:54 nbd: http://paste.uni.cc/15482 May 13 17:27:21 thepeople_mobile: working on it May 13 17:27:37 ok May 13 17:32:26 thepeople_mobile: http://nbd.name/mad.diff May 13 17:34:47 kaloz * r7212 /tags/kamikaze_pre1/ (scripts/arm-magic.sh target/linux/ixp4xx-2.6/config/default): disable not-fully supported ixp42x boards for pre1 May 13 18:20:50 thepeople_mobile: did the patch help? **** ENDING LOGGING AT Sun May 13 18:43:39 2007 **** BEGIN LOGGING AT Sun May 13 18:43:59 2007 May 13 19:09:49 nbd: trying now May 13 19:16:50 nbd: it builds fine for me (without your patch) May 13 19:18:46 it builds for me with the patch May 13 19:19:27 nbd * r7213 /trunk/package/madwifi/patches/100-kernel_cflags.patch: madwifi: fix xscale build May 13 19:20:10 loswillios: yeah, the bug was xscale specific :) May 13 19:21:55 i see. which device do you build on thepeople_mobile? May 13 19:24:22 loswillios: I build on FC6 May 13 19:27:10 is the a way to pass to $(eval $(call RequireCommand,/path/to/prog, two paths, found out that the prog is in different dirs on gentoo and debian sid (iirc) May 13 19:27:26 crazy_imp: just leave the path out May 13 19:28:01 thepeople_mobile: I was more referring to the HW May 13 19:28:17 nbd: ? only: $(eval $(call RequireCommand,prog, ? May 13 19:28:43 yes May 13 19:28:57 ok, thanks :) May 13 19:29:50 thepeople_mobile: cause xscale is not the usual consumer HW May 13 19:34:00 loswillios: I don't run any xscale hardware currently, but I compile images for the x-wrt repo May 13 21:56:34 florian * r7214 /trunk/target/linux/adm5120-2.6/ (3 files in 3 dirs): Fix routeboard board detection, add some new ZyXEL boards, thanks to Gabor Juhos May 13 21:58:07 florian * r7215 /trunk/package/base-files/files/usr/share/udhcpc/default.script: Do not suppress routes if the router received on renew is the same (#138) May 13 22:09:12 florian * r7216 /tags/kamikaze_pre1/package/base-files/files/usr/share/udhcpc/default.script: Backport udhcpc fix to kamikaze_pre1 May 13 23:03:34 nbd * r7217 /trunk/package/madwifi/patches/ (120-soc_fix.patch 300-napi_polling.patch): oops. some code accidentally slipped into the wrong patch file :) May 13 23:04:26 nbd * r7218 /trunk/package/madwifi/Makefile: madwifi: remove V=1 (was only added for debugging) May 13 23:08:52 ryd * r7219 /packages/utils/mksh/Makefile: added mksh to /etc/shells (#1685) **** ENDING LOGGING AT Mon May 14 02:59:57 2007