**** BEGIN LOGGING AT Tue Oct 16 03:00:02 2012 Oct 16 08:01:13 good morning Oct 16 08:18:43 how would i configure my denzil branches to only use udev and no systemd stuff ? Oct 16 08:32:28 rob_w: if you mean using meta-oe, probably not straightforward Oct 16 08:32:32 morning all Oct 16 08:36:34 bluelightning, yeah thx Oct 16 08:37:04 rob_w: since denzil we have done the split-out of meta-systemd and various other cleanups Oct 16 08:39:58 iam confused with meta-oe , poky/meta and oe-core some more ... Oct 16 08:40:39 i guess i could exlude meta-oe .. and only run poky denzil which looked like using a proper udev recipe .. Oct 16 08:41:17 where meta-oe denzil used newer udev version where there is now rcS.d udev script anymore Oct 16 08:42:32 yes, udev in meta-oe may well be broken for non-systemd cases unfortunately :/ Oct 16 08:43:02 you can set PREFERRED_VERSION_udev = "164" to stick with the OE-Core version at least Oct 16 08:43:27 did that before then then dropear misses systemd-SOMTHING Oct 16 08:43:48 maybe i need to use that version and some masks inside meta-oe ;-) Oct 16 08:44:06 what order of layers in my bblayer.conf do you recommend ? Oct 16 08:44:22 meta-oe before poky/meta ? or vize versa ? Oct 16 08:44:35 as i do need some apps from meta-oe Oct 16 08:45:10 I'd say meta-oe after; but the order is only going to affect how bbclass files are found Oct 16 08:45:58 isnt it the fact that if there similar recipes .. the later one will be the one ? hence overwrites ? Oct 16 08:46:49 like in hte old days with having a "overlay" Oct 16 08:49:39 rob_w: the layer priority affects recipe selection Oct 16 08:50:15 meta-oe has priority 6 and "meta" in OE-Core/Poky has 5, so recipes in meta-oe will overlay those in meta Oct 16 08:50:48 actually let's not start using "meta", let's just stick to OE-Core Oct 16 08:50:52 as a term Oct 16 08:54:13 ok nivr Oct 16 08:54:16 ok nice Oct 16 09:00:37 bluelightning, maybe you could enligth me about the probably simple facts on what is actuall a ?= doing Oct 16 09:01:13 rob_w: ?= sets the variable if not already set at the time the statement is parsed Oct 16 09:02:00 asummed .. but thx for clarify Oct 16 09:20:25 hi Oct 16 09:20:52 bluelightning: modphp 5.4.4 works with autotools do_configure - but have to fix apache headers Oct 16 09:20:53 hi hrw Oct 16 09:21:14 hrw: ah ok, great! Oct 16 09:21:19 hi florian Oct 16 09:21:55 hi bluelightning Oct 16 09:25:24 looks like it ignores apache headers path from axps Oct 16 09:32:13 bluelightning: apache2 apxs needs to return full path instead of /usr/include/apache2/ for headers Oct 16 09:32:43 hrw: I thought it already did... Oct 16 09:32:51 otherwise how was it working before? Oct 16 09:33:19 hrw: I thought it already did, otherwise how was it working before? Oct 16 09:37:46 network suxx today Oct 16 09:39:12 0: modphp-5.4.4-r1 do_package (pid 21437) Oct 16 09:48:04 hi hrw Oct 16 09:59:28 bluelightning: I see why it works for you probably. need to check Oct 16 10:08:40 bluelightning: fetch http://tygrysek.juszkiewicz.com.pl/~hrw/oe/ and try Oct 16 10:09:01 bluelightning: needs a bit of work still but works Oct 16 10:09:41 hrw: generally I prefer patches ;) Oct 16 10:09:46 but OK Oct 16 10:11:38 bluelightning: I moved modphp and php into one place and hacked it until it works. sharing as WIP Oct 16 10:12:03 now it is plain ugly Oct 16 10:12:18 still, downloading those files is much less easy than applying a single patch Oct 16 10:12:23 even if it is WIP Oct 16 10:13:22 ok, will generate patchset Oct 16 10:13:49 thanks... saves me trying to figure out how to wget recursively without jumping back up and downloading your entire website ;) Oct 16 10:14:09 wget -r -np :) Oct 16 10:14:27 or --recursive --no-parent Oct 16 10:14:30 ah, I hadn't scrolled down that far Oct 16 10:14:48 the wget manpage is useless :/ Oct 16 10:20:25 bluelightning: 4 patches for you Oct 16 10:26:12 bluelightning: I concur the wget manpage is useless I have to fight with it every time! Oct 16 10:26:47 hrw: thanks Oct 16 10:29:41 bluelightning: http://tygrysek.juszkiewicz.com.pl/~hrw/oe/0001-mod-php-next-set-of-changes.patch is 0005 Oct 16 10:31:10 0006 will follow - parseerrors Oct 16 10:36:07 and 0007 as well Oct 16 10:36:49 but first need to build Oct 16 10:50:34 hrw: ok, let me know when you're ready for me to grab them Oct 16 10:51:11 ok Oct 16 10:53:22 this rebuild should build Oct 16 11:33:30 hi guys Oct 16 11:35:24 I built my system with BB_GENERATE_MIRROR_TARBALLS enabled .... so I got git2_git.yoctoproject.org.prelink-cross.git.tar.gz ... I put that tar file in my downloads folder and enabled BB_NO_NETWORK = "1" so that it don't download prelink package Oct 16 11:35:42 ok, last rebuild of modphp Oct 16 11:37:10 I tried to build the prelink-native recipe and I got the following error http://paste2.org/p/2342191 Oct 16 11:37:29 can anybody help why prelink fails to fetch Oct 16 11:42:10 bluelightning: fetch 0007, apply and it should work Oct 16 11:44:41 bluelightning: 0008 adds forgotten dependency on apache to modphp Oct 16 11:53:59 auch... I did not had uptodate tree Oct 16 12:11:26 bluelightning: refreshed whole set Oct 16 12:12:16 ok I got it ... the particular commit was not included in tar file Oct 16 12:16:03 hrw: the moving would need to go the other way i.e. php in meta-oe should go to meta-webserver Oct 16 12:19:42 bluelightning: do as you want Oct 16 12:35:59 and we can update to 5.4.7 even Oct 16 12:50:05 hrw: so do you want me to tidy this series up or are you going to work on it further? Oct 16 12:51:09 bluelightning: please take a look at it as I may forget something there Oct 16 12:52:01 I can do migration meta-oe -> meta-webserver and 5.4.7 update after that Oct 16 12:52:10 hrw: I'm getting warnings about modphp trying to install duplicate files (into the native sysroot as well) Oct 16 12:52:49 headers probably? Oct 16 12:53:01 /home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/qemux86/usr/include/php/ext/mysqli/php_mysqli_structs.h Oct 16 12:54:39 hrw: http://pastebin.com/byJGq6pN Oct 16 12:55:11 actually I'm not clear on which recipe produced those Oct 16 12:56:00 bluelightning: both may Oct 16 12:56:21 well, neither should :) Oct 16 12:56:57 should those files even be installed at all? Oct 16 12:57:40 probably not Oct 16 13:00:53 hrw: btw, --without-mysql was deliberate, that's the old driver... also you've broken mysql PACKAGECONFIG by enabling pdo-mysql unconditionally Oct 16 13:01:21 I'm not opposed to enabling mysql by default but it should be able to be disabled still via PACKAGECONFIG Oct 16 13:03:48 ah, I know why... pdo-mysql was set for php. Oct 16 13:04:07 I don't like this SRC_URI_append-pn stuff either Oct 16 13:05:16 also taken from php Oct 16 13:06:29 I'd heard comments saying the php recipe needed cleaning up, I guess this is what those were referring to Oct 16 13:08:17 I do not know better way to not apply some patches for native targets Oct 16 13:10:37 i am using the lighttpd recipe (openembedded-core/meta/recipes-extended/lighttpd/lighttpd_1.4.30.bb) but i would like to change the configuration in my layer, the file is in openembedded-core/meta/recipes-extended/lighttpd/files/lighttpd.conf. Oct 16 13:10:39 what can i do ? Oct 16 13:11:27 do i have to make an empty recipe/bbappend Oct 16 13:11:53 or should i also redeclare file://lighttpd.conf ? Oct 16 13:12:07 afournier: add a bbappend with FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Oct 16 13:12:18 in your own layer Oct 16 13:12:21 arf Oct 16 13:12:38 ok Oct 16 13:12:49 then put your lighttpd.conf in a subdir called "lighttpd" Oct 16 13:13:14 you may also wish to add PRINC = "1" so that PR increases, that's not mandatory for your own private layer though Oct 16 13:13:43 if it's a good practise, let's do it Oct 16 13:14:13 is it normal the ":" at the end of the _prepend ? Oct 16 13:14:19 "${THISDIR}/${PN}:" Oct 16 13:17:10 bluelightning: re:php it does need cleaning up but it's not going to be a straight forward task I don't think Oct 16 13:23:52 jackmitchell: hmm, well I think it needs to be done before trying to merge the recipes together (or as part of the same exercise) Oct 16 13:23:59 I'm happy to help with that of course Oct 16 13:24:13 afournier: yes, paths in that variable are colon-separated Oct 16 13:24:35 ahhh Oct 16 13:24:36 ok Oct 16 13:26:59 bluelightning: I agree, no point in building on rickety foundations, I would love to help but it is a bit of a mine field Oct 16 13:27:15 I've tried before with little success Oct 16 13:27:48 I'm pretty sure I can at least tidy up the kind of nastiness I was referring to above Oct 16 13:28:19 heck, the complexity can't be any worse than the Qt recipes :) Oct 16 13:30:21 buildhistory can help you to check that the output hasn't changed when doing recipe cleanup as well Oct 16 13:30:38 morning all Oct 16 13:31:15 hi pb_ Oct 16 13:31:23 hi bluelightning Oct 16 13:31:31 bluelightning: haha I've never had to delve into the QT recipes just the snippets I see on the nailing list are enough to give me nightmares! Oct 16 13:32:03 s/nailing/mailing Oct 16 13:32:27 jackmitchell: they've improved since the OE-Classic days but still there are a lot of inc files involved, it's still a bit messy; I'm not clear on how it could be cleaned up much further though given what we need to build out of it Oct 16 13:33:44 bluelightning: I think it is the do install append and the ugly file variables which give it a particularity bad look Oct 16 13:34:25 bluelightning: there are also some QA warnings on building the recipe Oct 16 13:34:44 bluelightning: something else to go into the do_append rm batch Oct 16 13:34:52 hmm Oct 16 13:36:04 bluelightning: I went about trying to find all the different config options to strip the build down and remove some cruft but they're not particualrly well documented, or weren't a couple of months ago at lease Oct 16 13:37:24 jackmitchell: by all means if you find things that aren't documented in the yocto docs (at least, the upcoming ones for 1.3 which have been expanded and updated quite a bit), please file bugs Oct 16 13:44:59 bluelightning: ah sorry, I meant for stripping down the PHP configuration Oct 16 13:47:19 ah ok, fair enough Oct 16 13:47:39 you mean the php configure script options? Oct 16 13:53:52 bluelightning: yes Oct 16 13:54:50 ok, well my knowledge of those is almost certainly probably inferior to yours, I only know what I've picked up when working on the modphp recipe Oct 16 13:55:03 s/probably// Oct 16 14:23:44 morning Oct 16 15:20:46 it looks like libxml2 is linking with the host's zlib, rather than zlib-native Oct 16 15:20:55 how do I fix that? Oct 16 15:21:36 libxml2 shouldn't be linking against a native recipe, that'd be a problem :) Oct 16 15:21:42 i assume you mean rather than the zlib recipe Oct 16 15:22:15 ok, libxml2-native (built as a dependency of php-native) Oct 16 15:27:51 libxml2.inc says: Oct 16 15:27:56 DEPENDS_virtclass-native = "python-native" Oct 16 15:27:56 DEPENDS =+ "zlib" Oct 16 15:29:43 hollisb: I think you'll find libxml-native will still only have DEPENDS of "python-native" in that case Oct 16 15:30:20 hollisb: since the _virtclass-native override will just set the value Oct 16 15:30:49 yep, that'd explain it Oct 16 15:31:56 so when building a -native package, "DEPENDS =+ zlib" automatically becomes zlib-native? Oct 16 15:35:09 there's some EXTRA_OECONF stuff in libxml2.inc, and Google suggests adding "--with-zlib=${STAGING_LIBDIR}/=" might help? Oct 16 15:37:45 except without that last = of course :) Oct 16 15:42:34 bluelightning just explained the problem Oct 16 15:42:52 DEPENDS_virtclass-native replaces DEPENDS when that conditional is true, so the only native deps are python-native Oct 16 15:44:04 ah Oct 16 15:44:31 well, I had previously tried adding zlib-native there, but I'll give it another try Oct 16 15:45:12 yeah, in fact I have it right now and it's not working Oct 16 15:45:46 just change it from DEPENDS_virtclass-native to DEPENDS_append_virtclass-native Oct 16 15:45:50 problem solved Oct 16 15:45:53 ok Oct 16 15:45:59 with a leading space, obviously Oct 16 15:46:32 if it still doesn't work, then yeah, --with-zlib might help, assumign libxml2 has one Oct 16 15:46:47 it seems to, but it didn't help Oct 16 15:46:57 I'll reverify Oct 16 15:49:30 hmm Oct 16 15:58:51 I'm doing "bitbake -c clean libxml2-native; bitbake libxml2-native"; is that sufficient? the latter is completing suspiciously quickly Oct 16 16:08:49 is there a recipe for the mtdev library for using multitouch? Oct 16 16:11:53 hollisb: it will complete quickly if it's coming from sstate. -c clean doesn't guarantee a build from scratch :) Oct 16 16:18:18 kergoth: oh. well I guess this is a dumb question, but I want to force a rebuild (including do_configure). how should I do that, if "clean" doesn't get it done? Oct 16 16:22:58 hollisb: cleansstate = clean + remove sstate, thereby forcing a build from scratch. it's not often needed, however, as changes to the metadata change the signatures and therefore will result in the old sstate not being used Oct 16 16:23:51 oh, ok. just wanted to make sure I wasn't reporting false negatives Oct 16 16:27:27 kergoth: still broken with the patch at http://fpaste.org/tXOf/ Oct 16 16:27:49 er Oct 16 16:28:03 http://fpaste.org/ZaSh/ Oct 16 16:29:13 hollisb: you should be able to inspect the value of the variable using: bitbake -e libxml2-native | grep ^DEPENDS= Oct 16 16:29:20 saves you having to actually run the build Oct 16 16:30:21 bluelightning: cool. so with the patch in that paste, I get DEPENDS="pkgconfig-native autoconf-native automake-native libtool-native gnu-config-native zlib-native python-native python-native" Oct 16 16:30:34 and I can see the --with-zlib option being used for configure Oct 16 16:30:58 but somehow the end libxml2.so is still using the host's /lib/libz.so Oct 16 16:54:43 I see openembedded-devel@lists.openembedded.org is subscriber-only huh? a little hostile Oct 16 17:00:27 hollisb, I assume you mean it does not forward emails from non-subscribers? I would assume this is reduce the amount of spam Oct 16 17:01:13 I would assume so too, but it's rather heavy-handed... lots of lists manage spam without rejecting outside contributions Oct 16 17:02:14 I did not setup that list, but lists I mange hold non-subscribers for moderation Oct 16 17:02:19 and 99% of that is spam Oct 16 17:02:36 yup, I do the same Oct 16 17:03:05 of course, you only need to moderate the very first post, because you can add them to the "always allow" list after that Oct 16 17:03:20 so by definition, most moderation requests are spam :) Oct 16 17:36:44 I see COMPATIBLE_MACHINE everywhere. Is there an inverse, so a can say "don't use this version on this machine" ? Oct 16 17:39:58 awozniak: COMPATIBLE_MACHINE_yourmachine = "something" Oct 16 17:40:05 works fine Oct 16 17:40:28 hrw: what's "something" ? Oct 16 17:40:40 awozniak: any value Oct 16 17:40:50 awozniak: you can write reason there etc Oct 16 17:41:05 anything other then yourmachine Oct 16 17:41:09 hrw: and that will signal that the recipe is NOT compatible on yourmachine ? Oct 16 17:41:28 yes Oct 16 17:41:32 ah, I see Oct 16 17:41:35 Thanks Oct 16 17:41:36 in kind of reverse way Oct 16 17:42:22 Dear OE, for MACHINE=mymachine please consider COMPATIBLE_MACHINE != "mymachine" and do not build this recipe. Oct 16 17:43:03 COMPATIBLE_MACHINE += "!mymachine" Oct 16 17:43:19 no Oct 16 17:43:38 or.. maybe it would work even Oct 16 17:43:46 (was a suggestion, != is never going to fly) Oct 16 19:47:38 kergoth: any further ideas on the libxml/zlib problem? **** ENDING LOGGING AT Wed Oct 17 03:00:01 2012