**** BEGIN LOGGING AT Wed Aug 10 02:59:57 2011 Aug 10 03:38:09 zeddii, ping Aug 10 15:17:59 * zeddii simply cannot build perl for qemumips Aug 10 15:22:46 pastebin? Aug 10 15:40:28 http://pastebin.com/5jkZMsyV Aug 10 15:40:44 I had just given up trying to smash it into submission. Aug 10 15:40:54 I just built qemuarm on that very same machine, no issues. Aug 10 15:41:16 I've smashed sstate-cache, cleanall, perl-native, perl, everything I could think of. Aug 10 16:19:13 zeddii: have you tried with complete build from scratch (not from sstate) Aug 10 16:20:08 khem. not two seconds ago I move aside my entire sstate and tmp. will try that build. Aug 10 16:20:30 I had just done that for qemuarm, which was the only other build present, so what's another one!? :) Aug 10 16:21:23 zeddii: I have seen that error sometimes Aug 10 16:21:40 but I usually nuke my build once a week Aug 10 16:21:45 fun of doing toolchains Aug 10 16:22:07 khem: ah yes. I was a toolchain guy about 8 years ago. I feel the pain! Aug 10 16:22:16 but strangely .. that does make me feel 'better' :) Aug 10 16:22:57 heh Aug 10 16:23:29 I am hacking on llvm these days Aug 10 16:23:36 not in OE context yet Aug 10 16:24:10 cool. I sat in on the llvm talk @ ELC Aug 10 16:25:31 but its no more than arm and x86 atm Aug 10 16:25:49 but should be an appealing alternative for gcc soon Aug 10 16:25:52 a start though .. Aug 10 16:26:02 more of the egcs game from years ago :P Aug 10 16:26:11 its bigger than that Aug 10 16:26:18 egcs was more of a fork Aug 10 16:26:25 this is a knife :) Aug 10 16:26:30 agreed. this is hardly related. Aug 10 16:26:58 I like clang's diagnostics Aug 10 16:27:07 and the way its coded is clean too Aug 10 16:28:03 nice. it did sort of start out academic, but is making the leap to useful :) Aug 10 16:28:07 but if it has to get any traction it has to support gnu'ism may be for a little while to port over the apps Aug 10 16:28:18 yes Aug 10 16:28:23 that's the bane of soo many compilers. Aug 10 16:28:28 ICC, diab, etc, etc. Aug 10 16:28:37 everyone asks "but can you build the kernel?" ;) Aug 10 16:28:39 code generation is at par with gcc on arm and x86/64 Aug 10 16:28:46 with all that arcane gnu-foo. Aug 10 16:29:11 Intel compiler is the best among others in supporting gnu'ism Aug 10 16:29:34 it has come a long way (or so I've heard), haven't had the chance in a while to use it. Aug 10 16:30:08 gcc wins with multiarchitectural support Aug 10 16:30:11 otherwise it sucks Aug 10 16:32:59 wide .. but not deep! :) Aug 10 16:33:20 * zeddii steps out. -EFOOD Aug 10 20:11:02 ERROR: Function 'File: '/home/chinmaya/yocto/poky-5.0.1-build/downloads/gnutls-2.10.4.tar.bz2' has md5 checksum b3a1fe8fae9fbeba7f88484da281ecfc when 4e1517084018a8b1fdc96daabea40528 was expected (from URL: 'ftp://ftp.gnutls.org/pub/gnutls/gnutls-2.10.4.tar.bz2')' failed Aug 10 20:11:02 NOTE: package gnutls-2.10.4-r0: task do_fetch: Failed Aug 10 20:11:24 Why is it coming like this?? Aug 10 20:12:27 either the download was partial (which can happen).. or someone has changed the original file upstream which changes the md5sum.. Aug 10 20:12:46 I would suggest you run a bitbake -c cleanall gnutls and try it again.. Aug 10 20:13:08 if it still fails, look at /home/chinmaya/yocto/poky-5.0.1-build/downloads/gnutls-2.10.4.tar.bz2 and see if it is the file you were expecting Aug 10 20:13:20 (and not a text chunk that says file not found or something like that) Aug 10 20:13:36 Thanks for the reply.. yes it is saying file not found Aug 10 20:13:51 when i tried to $rm it Aug 10 20:14:33 "$ bitbake -c cleanall gnutls" --> is this you want me to run Aug 10 20:15:23 running bitbake -c cleanall gnutls should remove the file from your ownloads directory Aug 10 20:15:58 then running your previous bitbake command will attempt to download it. if the download fails then the problem was likely a partial download.. if the download succeeds and you get the same md5sum error.. check thats the file is intact.. Aug 10 20:16:54 Thanks a lot.. ohk.. I'm running it now... Aug 10 20:39:30 afternoon all Aug 10 20:43:39 fray, Thanks a lot again ! It worked like a charm :) Aug 10 21:05:02 fray: ping Aug 10 21:05:18 I'm here Aug 10 21:05:39 fray: just wanted to check with you on a couple of bugs relative to M3 Aug 10 21:05:53 fray: 1352 and the pseudo bug Aug 10 21:05:55 ok Aug 10 21:06:05 I'm still working on 1352 Aug 10 21:06:15 pseudo bug, I need someone to reproduce it with a debugger attached.. Aug 10 21:06:27 It really doesn't look like pseudo fromt he description.. but I haven't yet had time to do it myself Aug 10 21:06:36 (it looks like a python, sqlite, or similar) Aug 10 21:06:48 but I need to know what the system is doing, cause it might be bad data from pseudo Aug 10 21:07:09 fray: hmm, ok, problem is it's intermittent enough that I am not sure how we are going to get there. OK, I will bang away at that one some also. Aug 10 21:08:35 Jessica said that she thought it was happening often enought o be reproduced Aug 10 21:08:47 (I'm referring to 1332) Aug 10 21:09:07 Ok, I will check with here, I have only seen it twice on the builds happening here. Aug 10 21:09:52 ok Aug 10 21:10:14 BTW, just catching up and maybe this is already been pulled into oe-core, did you see the patch from jani on the yocto list adding the magic file option to package_rpm? Does that look reasonable to you? Aug 10 21:10:41 let me look, I did look at it, but I don't remember Aug 10 21:11:27 I'm not finding it.. what was the topic? Aug 10 21:13:10 ya, I can't find it at all Aug 10 21:16:35 fray: Re: [yocto] Magic/File problems Aug 10 21:17:18 thats why.. I was looking at the wrong list Aug 10 21:24:40 BTW thats about 19 liters.. easily 20 when you top it off.. Aug 10 21:24:50 oops Aug 10 22:15:43 sgw -- sorry got interrupted.. the magic patch should be just fine.. Aug 10 22:25:18 hm.. how can one toggle CCACHE off? Setting CCACHE = "" in local.conf seems not enough Aug 10 22:29:23 how do I define a var if another is defined? Aug 10 22:29:27 i want to do something liek Aug 10 22:29:34 FOO = blah bitbake world Aug 10 22:29:56 and have BAR = blah iff FOO is defined Aug 10 22:30:17 looking at base_contains, and base_conditional Aug 10 22:30:22 you'd have to use python for that. anonymous python function, ${@}, or an event handler, depending on what you're trying to acheive Aug 10 22:30:45 python is fine.. i was thinking something might be defined... Aug 10 22:31:13 base_conditional should do Aug 10 22:31:49 lets see.. Aug 10 22:31:58 i tried BAR = ${@base_conditional ("FOO", "", "default_value", "${FOO}")} Aug 10 22:32:17 ah, your issue there is that an undefined var returns None, not the empty string Aug 10 22:32:19 and I got failure expanding variable name None Aug 10 22:32:20 either do a FOO ?= "" Aug 10 22:32:31 ahh Aug 10 22:32:34 fray: thanks Aug 10 22:32:36 or check for None instead of "" Aug 10 22:36:56 seems to be working nice.. thanks =) Aug 10 22:58:22 Anyone have any thoughts on automatically adding more context to our log records for messages initiated by the task subprocesses? I'm thinking at hte very least filename and task id and task name. would make it nicer for the UIs to associate messages to tasks directly rather than using the process ids, I think, and could uatomatically ensure that all warnings/errors/etc coming from tasks include identifying information about which task it Aug 10 22:58:22 came from, for eg. knotty Aug 10 22:58:41 wouldn't think it'd be terribly difficult to do, just add a filter or two after forking the task process Aug 10 23:01:52 msm: np **** ENDING LOGGING AT Thu Aug 11 02:59:56 2011