**** BEGIN LOGGING AT Thu Apr 24 02:59:58 2014
Apr 24 08:42:51 <bluelightning> morning all
Apr 24 09:02:32 <afournier> morning
Apr 24 10:43:16 <afournier> Parsed BBPATH is['', '/tmp/OE/openembedded-core/meta', ...
Apr 24 10:43:29 <afournier> how can i find where this empty path is added ?
Apr 24 10:53:04 <afournier> i got it, it's because the first layer is oe-core/meta BBPATH starts with a ':' that is later plit in python so the first item in the array is '' :(
Apr 24 10:57:05 <afournier> defining bbpath in bblayers fix the problem... weird
Apr 24 11:07:50 <Jin^eLD> hi
Apr 24 11:11:13 <Jin^eLD> if a .bb does a do_install_append() and I have a .bbappend where I want to append yet more stuff, my do_install_append() would overwrite the append in the original .bb.... is there a way on how I could append_append or something, if you know what I mean? :)
Apr 24 11:11:34 <Crofton|work> This doesn't do what I want
Apr 24 11:11:37 <Crofton|work> DEPENDS = "bzip2 mcpp expat openssl python db mcpp"
Apr 24 11:11:38 <Crofton|work> DEPENDS_prepend_class-target = "zeroc-ice-native"
Apr 24 11:12:02 <bluelightning> Crofton|work: missing trailing space?
Apr 24 11:12:19 <Crofton|work> in the append?
Apr 24 11:12:26 <bluelightning> Jin^eLD: it won't overwrite it, it'll append
Apr 24 11:12:29 <bluelightning> Crofton|work: the prepend
Apr 24 11:12:47 <Crofton|work> sorry fliiping betweeb append and prepend
Apr 24 11:12:49 <Jin^eLD> bluelightning: hmm.. ok let me do a -e and check, I had the feeling it was overwriting...
Apr 24 11:13:17 <bluelightning> Crofton|work: if it was an _append it'll need to be a leading space
Apr 24 11:13:23 <Crofton|work> right
Apr 24 11:13:26 <Crofton|work> thi shelps
Apr 24 11:20:17 <Jin^eLD> bluelightning: ok thanks, you are right, it really is appended, then my problem is somewhere else :)
Apr 24 11:28:45 <Crofton|work> bluelightning, will stick with prepend and trailing space. Keeps all the tricks on one line
Apr 24 14:16:28 <heeen> does oe have a feature to insert source revision as a string into code
Apr 24 14:16:57 <heeen> or build date
Apr 24 14:40:53 <WarheadsSE> ? for any -- any way to sanely get -u goggle while getting knotty log?
Apr 24 14:41:42 <WarheadsSE> We're using knotty | tee, but some user exerpience issues, such as ctrl-c not getting passed to bitbake
Apr 24 14:42:27 <kergoth> ^C works pretty well for me with bitbake. also goggle doesn't even work anymore with master afaict. i'm guessing you're on danny or so?
Apr 24 14:44:49 <WarheadsSE> no, we're in dora & dylan
Apr 24 14:45:25 <WarheadsSE> kergoth: this all goes tied back to dockbake && bakethat wrapper
Apr 24 14:46:06 <WarheadsSE> in the -i interactive mode, ctl-c doesn't stop the bugger
Apr 24 14:47:27 <kergoth> ah, fix the wrapper's signal handling :)
Apr 24 14:47:53 <WarheadsSE> 99% likely, is what I am thinking
Apr 24 14:49:58 <bluelightning> kergoth: hmm, if goggle is broken, is there a bug open for that?
Apr 24 14:50:45 <kergoth> not sure, i was planning on opening one. but honestly most of the uis are broken
Apr 24 14:50:48 <kergoth> ncurses, etc
Apr 24 14:50:56 <kergoth> should be removed if not fixed
Apr 24 14:51:40 <kergoth> WarheadsSE: check out http://www.cons.org/cracauer/sigint.html, i found the info there to be quite helpful
Apr 24 14:53:19 <kergoth> iirc even the oe-core bitbake wrapper wasn't handling the signals properly either :)
Apr 24 14:53:31 <pompomJuice> Anyone here get the following auto tools error when building Mono 3.4 in oe http://pastebin.com/6HJMTNkE
Apr 24 14:54:08 <pompomJuice> it is not so much a mono bug, than it is a autotools bug
Apr 24 14:54:12 <pompomJuice> I am not sure
Apr 24 14:55:29 <pompomJuice> basically, mono has a autogen.sh that calls ./configure in subfolders. One of these ./configures generates a file called config.status
Apr 24 14:56:13 <pompomJuice> this config.status differs from one generated normally (withouth bitbake, normal linux build)
Apr 24 14:56:31 <pompomJuice> the bitbake one is generated when compiling the native mono
Apr 24 14:56:46 <pompomJuice> meaning its mono-native building, so they should be the same
Apr 24 14:57:24 <pompomJuice> the actuall difference is very small
Apr 24 14:57:59 <pompomJuice> in one file a variable comes out as this     ofile='-libtool'
Apr 24 14:58:07 <pompomJuice> the bitbake version
Apr 24 14:58:16 <pompomJuice> and the normal version looks like this ->     ofile='libtool'
Apr 24 14:58:22 <pompomJuice> there is a dash in there
Apr 24 14:58:54 <pompomJuice> which causes a later command $RM "$cfgfile" to fail because it resolves to $rm -f -libtoolT
Apr 24 15:01:30 <pompomJuice> This does not happen with mono 3.2.8
Apr 24 15:04:45 <pompomJuice> the problem is deep inside our beloved autotools
Apr 24 15:05:12 <pompomJuice> nobody really knows how autotools work
Apr 24 15:05:43 <Crofton|work> :)
Apr 24 15:05:46 <Crofton|work> very true
Apr 24 15:05:51 <Crofton|work> fray, https://bugzilla.yoctoproject.org/show_bug.cgi?id=6212
Apr 24 15:06:09 <kergoth> when doing builds out of oe, our macros cause the emitted libtool script to be prefixed by the target_sys
Apr 24 15:06:23 <kergoth> i'm guessing thats how you end up with '-libtool', the var it used to prefix was undefined?
Apr 24 15:06:24 * kergoth shrugs
Apr 24 15:06:31 <Crofton|work> letothe2nd says deb and ipk are fine, seems like rpm is writing somewhere outside a chroot
Apr 24 15:06:36 <pompomJuice> mmm
Apr 24 15:06:42 <pompomJuice> I am not as clued up
Apr 24 15:06:43 <pompomJuice> give me a sec
Apr 24 15:06:48 <pompomJuice> to check 
Apr 24 15:06:54 <kergoth> that reminds me, i'm sick of seeing ubinize.cfg ending up in $PWD/$TOPDIR
Apr 24 15:07:45 <pompomJuice> prefix
Apr 24 15:07:52 <pompomJuice> my prefix is dodgy
Apr 24 15:08:26 <pompomJuice> when you build mono from git, I have my configure step as ./autogen.sh --prefix=${D}/usr
Apr 24 15:08:35 <pompomJuice> that could be it
Apr 24 15:08:37 <pompomJuice> let me permutate
Apr 24 15:09:31 <kergoth> no, you're confused
Apr 24 15:09:37 <kergoth> i said the prefix of the libtool script name
Apr 24 15:09:40 <kergoth> not the variable name 'prefix'
Apr 24 15:09:44 <pompomJuice> thats possible, even probable
Apr 24 15:09:44 <kergoth> which is completely and totally different
Apr 24 15:10:05 <kergoth> the 'prefix' variabl ename is a path on the target. the libtool script is prefixed by target sys, as i said before
Apr 24 15:10:31 <pompomJuice> I am sorry am am to noob to undestand what you are saying
Apr 24 15:10:51 <kergoth> then it's unlikely you'll be able to fix this without learning a great deal more about autotools
Apr 24 15:11:06 <kergoth> i'd suggest contacting the mailing list or opening a bug, or reading more about autotools :)
Apr 24 15:11:17 <pompomJuice> makes sense
Apr 24 15:11:33 <pompomJuice> ok but one more try
Apr 24 15:11:39 <pompomJuice> maybe you can guess
Apr 24 15:11:46 <pompomJuice> basically
Apr 24 15:11:51 <pompomJuice> building mono from git
Apr 24 15:12:04 <pompomJuice> ./autogen --prefix=/usr
Apr 24 15:12:07 <pompomJuice> make
Apr 24 15:12:08 <pompomJuice> make install
Apr 24 15:12:20 <pompomJuice> so when I make my recipy for bitbake
Apr 24 15:12:31 <pompomJuice> the mono-native contains one modification
Apr 24 15:13:05 <pompomJuice> the do_configure is overwritten with a new one containing the ./autogen.sh --prefix=${D}/usr
Apr 24 15:13:08 <pompomJuice> that is it
Apr 24 15:13:17 <pompomJuice> somehow that comes out different
Apr 24 15:13:37 <pompomJuice> even though according to what I know its a native build so it should be the same, unless there is a export changing the behaviour
Apr 24 15:13:45 <pompomJuice> anywy
Apr 24 15:13:50 <pompomJuice> thanks for helping
Apr 24 15:14:01 <pompomJuice> at least I know its not a noob mistake somewhere
Apr 24 15:14:24 <pompomJuice> further more
Apr 24 15:14:30 <pompomJuice> building mono 3.2.8 from git worked
Apr 24 15:14:39 <pompomJuice> so it is something on the 3.4 branch that is causing this
Apr 24 15:17:13 <pompomJuice> I have a new plan
Apr 24 15:17:44 <pompomJuice> figure out once and for all what the hell autotools do
Apr 24 15:18:41 <pompomJuice> as on a bash scripting level, because knowing what autotools do on a conceptial level is simply not enough
Apr 24 15:18:57 <pompomJuice> see you guys in 50 years
Apr 24 15:19:15 <kergoth> i'd recommend getting at least a high level handle on the m4 language, as that's what autoconf configure.{in,ac} scripts and .m4 files are written in
Apr 24 15:19:23 <kergoth> there are books on autotools that might be of use
Apr 24 15:19:24 <pompomJuice> perfect
Apr 24 15:19:25 <pompomJuice> thanks
Apr 24 15:19:28 <pompomJuice> I did not know that
Apr 24 15:19:31 <pompomJuice> il start there
Apr 24 15:19:41 <kergoth> see also the manuals at gnu.org. they're surprisingly good
Apr 24 15:19:46 <kergoth> libtool, autoconf, automake, and even make
Apr 24 15:20:01 <pompomJuice> to many times now have I run into autoconf/automake/configure bugs that are just greek
Apr 24 15:20:03 <kergoth> they don't really help you see how the pieces fit together, but for docs on each piece, there's value
Apr 24 15:20:25 <pompomJuice> ok but its 5pm
Apr 24 15:20:28 <pompomJuice> going home
Apr 24 15:20:29 <kergoth> https://www.sourceware.org/autobook/autobook/autobook.html
Apr 24 15:20:31 <kergoth> later
Apr 24 15:20:33 <pompomJuice> studying m4 tonight
Apr 24 15:20:39 <pompomJuice> and that
Apr 24 15:20:51 <pompomJuice> thanks
Apr 24 15:20:54 <kergoth> np
Apr 24 15:21:43 <nerdboy> anyone still wants to go to oedam and cost-share the room, i'd be up for that...
Apr 24 15:22:22 <nerdboy> go one reserved at a slightly cheaper hotel a couple blocks over
Apr 24 15:23:11 <nerdboy> or not...
Apr 24 15:24:54 <nerdboy> just as an aside, the weirdest thing i learned doing that with another gentoo dev was that french developers like colorful underwear...
Apr 24 15:26:39 <WarheadsSE> kergoth: yeah, nothing like indeterminate behavior :P
Apr 24 15:27:47 <WarheadsSE> So far, simple exit trap works unless bitbake has made it to the middle of parsing recipes
Apr 24 15:36:26 <WarheadsSE> Additionally, it seems the more BBTRHEADS the longer it takes to populate, by a large amount.
Apr 24 15:47:29 <kergoth> WarheadsSE: when I've scripted bitbake, i've often backgrounded it, intercepted the user's signals, and then killed bitbake when received. not ideal, though, since you don't get interactive knotty when doing it that way
Apr 24 15:48:45 <WarheadsSE> yeah, I have afeeling my description beteween knotty/goggle is incorrect by my own understaing gaff, and it might just be the bahviro difference between where fd1 is pointed (tee, &'d)
Apr 24 19:10:18 * ajtag is away: Away
Apr 24 19:10:37 * ajtag is back (gone 00:00:03)
**** ENDING LOGGING AT Fri Apr 25 02:59:58 2014