**** BEGIN LOGGING AT Mon Jan 13 02:59:59 2014 Jan 13 06:31:43 i'm using sstate-cache Jan 13 06:31:54 but i'm getting error Jan 13 06:32:05 fatal error: stdio.h: No such file or directory Jan 13 06:32:17 after using sstate Jan 13 08:02:35 JaMa: Good morning. Do you remember why this revert was done? (sry if I asked twice -- needed to run last time) https://github.com/openembedded/meta-oe/commit/c8aa4b10b2ab341bbcb163d041c44c1a20172ee1#diff-e738151ff5956440b2b29ff001e00311 Jan 13 08:03:14 JaMa: reverts nodejs 0.10.x Jan 13 08:49:08 likewise: Hi, it was pushed by accident and wasn't ready (gateone isn't compatible with that) Jan 13 08:49:28 likewise: if you're intested in nodejs 0.10.x you can check meta-webos layer Jan 13 09:19:43 JaMa: thanks Jan 13 09:25:45 I want to make a recipe that installs a config file and use that config by another recipe in the do_configure stage. I tried a number of things, but without success. How do I do this? Jan 13 09:33:07 where should a gptfisk recipe go? Jan 13 09:33:12 oe-core or meta-oe? Jan 13 09:33:19 and if meta-oe, which layer? Jan 13 09:36:31 gptfdisk you mean? Jan 13 09:40:56 ant_work: yes Jan 13 09:41:12 * koen can't figure out parted Jan 13 09:41:27 and parteds i18n is &*#%*%# annoying Jan 13 09:41:50 koen: heh, last days I've hackintoshed again ;) Jan 13 09:42:03 morning all Jan 13 09:44:04 hey bluelightning Jan 13 09:44:13 hi koen Jan 13 09:44:35 ant_work: I've been doing winter cleaning on my x86 machines Jan 13 09:44:58 my hatred for ACPI has increased tenfold Jan 13 09:45:29 or rather vendor provided ACPI blobs that are $*($*(@ broken Jan 13 09:46:01 and of course you can buy x86 systems with no ACPI in their BIOS Jan 13 09:46:14 but intel in all their wisdom only supports ACPI based cpufreq on atoms Jan 13 09:47:51 ah, the lovely ACPI tables ;) Jan 13 09:49:28 why load a proper linux driver when you can just bitbang it using bytecode? Jan 13 09:49:39 ant_work: I have no problem with the tables Jan 13 09:49:52 you need a way to describe HW Jan 13 09:53:16 but I do have a new appreciation for UEFI and especially UEFI shell Jan 13 09:57:07 TobyG: You can copy that config file to sysroot and then you can use i think Jan 13 09:57:40 koen: nostalgy of bare-metal arm bootloaders? Jan 13 09:57:50 i'm using sstate-cache Jan 13 09:57:58 but i'm getting error Jan 13 09:58:06 fatal error: stdio.h: No such file or directory Jan 13 09:58:12 after using sstate Jan 13 09:58:27 koen: fwiw I was tempted to go back to lilo whan Gentoo switched to grub2 ;) Jan 13 09:58:46 ant_work: me too Jan 13 09:58:52 i have removed thhe build folder from where i am using the sstate Jan 13 09:59:08 ant_work: but I figured out that you don't need to use the byzantine grub-install Jan 13 09:59:15 just write your own config :) Jan 13 09:59:35 ant_work: my system still had grub 1.97, which didn't support ext4 Jan 13 10:00:02 koen: I'm beating grub2 with --force to install in the linux partitions... Jan 13 10:01:34 UEFI is so nice Jan 13 10:01:43 just use uefi stub :) Jan 13 10:01:49 or if you want gummiboot Jan 13 10:01:54 no need for grub Jan 13 10:02:25 but since I'm hypocritical I'm trying to get grub2 working on arm using uboot as chainload :) Jan 13 10:04:58 kexec -l grub Jan 13 10:10:18 kexec -l grub ? Jan 13 10:10:19 hmm Jan 13 10:12:05 u-boot -> linux -> grub -> linux would be a bit much Jan 13 10:14:22 just a lil bit Jan 13 10:22:04 when I find some time I'll create a kexecboot-USB dongle and try to boot onboard grub Jan 13 10:24:23 * likewise remembers discussion during bareboot talk where gleixner tried to convince the future bootloader for the Linux kernel... is a Linux kernel. (i.e. why duplicate more and more drivers into the bootloader). Jan 13 10:24:37 ^convince +that Jan 13 10:26:28 * likewise also remembers writing a Linux kernel into the BIOS PROM (coreboot predecessor on Via M 6000 ITX board). From power-on to prompt in <2 secs on x86... Jan 13 10:26:58 microsoft tried to fix that with their fastboot requirement Jan 13 10:27:19 iirc POST needs to be below 2s Jan 13 11:09:44 from zero to nowhere in under 2 secs with microsoft Jan 13 11:34:00 morning all Jan 13 11:37:15 pb: good morning Jan 13 12:58:16 morning pb_ Jan 13 13:17:32 hi bluelightning, likewise Jan 13 16:04:14 I'm getting an error when baking classpath-native: http://paste.debian.net/75948/ Has anybody seen this? Jan 13 16:05:42 I'm using meta-java, master branch. Jan 13 16:06:58 mario-goulart: have you updated recently? Jan 13 16:07:09 mario-goulart: there were some recent commits to fix this Jan 13 16:07:40 jackmitchell: I updated the meta-java tree yesterday. Jan 13 16:08:10 mario-goulart: I would post with a full config log in here then: https://github.com/woglinde/meta-java/issues/49 Jan 13 16:13:36 jackmitchell: thanks. That thread mentions a patch I don't have in my tree (https://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg34182.html). It seems that it is not merged into meta-java yet. Jan 13 16:26:11 I wonder where that PR bump comes from in the patch. Master doesn't even have PR and in the openjdk-7 branch, PR is 3. Jan 13 16:31:17 mario-goulart: yeah, I'm not sure that patch is relevant Jan 13 16:31:36 mario-goulart: I would just make sure you're on meta-java/master Jan 13 16:32:03 the openjdk-7 branch is _old_ Jan 13 16:32:10 jackmitchell: yeah, I'm using master. Jan 13 17:39:11 If I try to bake jikes, do_configure fails because jikes-initial (a shell script) can't find jikes. The actual error message is in cacao-initial's config.log. Looks there's some dependency issue somewhere. Jan 13 18:46:40 hi, are PV sorted like debian versions are sorted? e.g. I am especially interested to know if we have 1.0~1 < 1.0 < 1.0+1 Jan 13 19:31:15 ndec: iirc it's not completely like debian Jan 13 19:31:26 ndec: in oe-classic we had an ipkg-version-compare script Jan 13 19:31:45 ndec: but then opkg was fixed upstream, so the script didn't match opkg anymore :( Jan 13 19:34:17 and it has C and shell version of ipkg-version-compare in opkg-utils Jan 13 19:36:01 does opkg-make-index match opkg.c nowadays? Jan 13 19:36:38 I remember opkg-make-index doing the right thing and hence masking bugs in opkg Jan 13 19:50:36 he also asked about PV specifically, so it's possible he's asking about recipe version comparison for the default preferred version, which of course is the comparison in the bitbake code rather thant he bits in opkg-utils Jan 13 19:50:38 * kergoth shrugs Jan 13 19:57:49 It seems that the problem with the java stuff boils down to "cacao-initial: resolve.c:348: resolve_classref_or_classinfo: Assertion `c->state & 0x0002' failed." Jan 13 19:57:57 sorry for asking a common question: How can I determine why the hash has changed on a shared state file? Jan 13 19:58:26 I've seen that error mentioned on this channel at least two times in 2009, but no solution. Jan 13 20:08:25 koen: kergoth: JaMa: sorry, was having dinner... so, yes it's about PV. i am currently working on recipes that will be v2.0 soon, but aren't yet. so I wanted to use 2.0~svnREV for the version, so that when v2.0 comes out it is higher. Jan 13 20:08:53 ndec: just set DEFAULT_PREFERENCE = "-1" in the pre-release recipe, so it will never be autoselected unless the user specifically chooses it with PREFERRED_VERSION Jan 13 20:09:06 then you dont' have to worry about the peculiarities of the sorting, unless you're setting up package feeds Jan 13 20:10:06 hmm.ok. Jan 13 20:11:14 never autoselected, what happens if there is only 1 recipe and it has DEFAULT_PREFERENCE=-1? Jan 13 20:17:42 What goes into an OEBasicHash? Jan 13 20:25:34 koen: I don't remember, but IIRC I was fixing one of them to do the same as the other one Jan 13 20:26:34 kergoth: or unless they are in different layer Jan 13 21:05:08 ndec: if there's only one eligible recipe then it will be selected no matter how low its preference is. Jan 13 21:05:25 ah, thanks. Jan 13 21:06:39 morning kergoth Jan 13 21:19:55 hi. is it considered 'ok' to do something like RPROVIDES_${PN} = 'virtual/foobar' in a recipe? i would have expected so, but couldn't find such a pattern anywhere...so i am wondering if i should do that.. Jan 13 21:20:37 basically, i need to include an implementation (and only 1) of 'foobar' in an image. and I have multiple 'provider' of foobar. Jan 13 21:20:41 ndec: no, virtual/* are usually used only for build-time dependencies Jan 13 21:21:03 so, how do we do that for run time depends? Jan 13 21:21:25 ndec: for runtime it doesn't work very well, you can use RPROVIDES_${PN} = 'foo', but to determinisc selection of runtime providers use VIRTUAL-RUNTIME variable Jan 13 21:21:47 VIRTUAL-RUNTIME_foo ?= "bar" Jan 13 21:21:59 RDEPENDS_${PN} = "${VIRTUAL-RUNTIME_foo}" Jan 13 21:22:16 and then you can change VIRTUAL-RUNTIME_foo = "something-else" in your distro config Jan 13 21:22:18 ah. cool. Jan 13 21:22:58 that's exactly what i need. i didn't know about this var.. Jan 13 21:24:12 it's not really a variable, just a convention Jan 13 21:24:29 yeah, that's what i just realized.. Jan 13 21:24:59 so, why not using virtual/xxx for run time? any particular reasons? Jan 13 21:26:42 package managers don't really support it properly Jan 13 21:27:09 AIUI Jan 13 21:34:24 bluelightning: JaMa: thanks. and it just works ;-) Jan 13 21:43:12 ndec: "virtual/xx" wouldn't work for runtime because "/" isn't a valid character in package names :-} I think there is some precedent for RPROVIDES_.. = "virtual-runtime-xx" but possibly not in oe-core itself. Jan 13 21:44:15 but, like bluelightning said, package manager support is fairly patchy. opkg in particular is particularly bad at dealing with semi-virtual packages. Jan 13 21:44:25 ok. that explains it. too bad it was defined as a convention with a '/' ;-) Jan 13 21:44:38 I think there was some other reason than invalid character Jan 13 21:45:38 http://lists.openembedded.org/pipermail/openembedded-core/2012-December/071540.html Jan 13 21:47:47 ah now I remember, the main problem is that PREFERRED_PROVIDER setting isn't respected by packagemanagers and people assume that RPROVIDES/RDEPENDS configuration work the same as build-time deps (P_P) **** ENDING LOGGING AT Tue Jan 14 03:00:00 2014