**** BEGIN LOGGING AT Fri Mar 11 02:59:58 2016 Mar 11 08:01:34 hello, I have a patch I want to contribute to yocto (meta/) but I can't figure out what branch to follow, is it master or master-next Mar 11 08:01:55 it seems master-next is regularly rebased, so I would guess master, but i couldn't find any doc Mar 11 08:22:46 boucman_work: master Mar 11 08:22:57 thx Mar 11 08:24:02 and do you follow the convention of using --- in the commit message to separate the actual message from the comment to the ML ? or do you rather have a [0/1] message where i discuss stuff about the patch ? Mar 11 08:36:22 boucman_work: i prefer cover letter if the patch requires more explanation than what's in the commit message Mar 11 08:47:37 boucman_work: [0/1] is for the discussion Mar 11 08:48:03 boucman_work: although the commit message should be self supporting Mar 11 08:55:40 hi everyone Good Morning, i build java 7 jre package with bitbake, i must to run this command: bitbake openjdk-7-jre, Am i right ? Mar 11 08:57:36 LetoThe2nd: i think LetoThe2nd exactly know this question ^_^ ? Mar 11 08:58:30 huh? Mar 11 08:58:44 why should i? Mar 11 08:59:26 LetoThe2nd: My question is : hi everyone Good Morning, i build java 7 jre package with bitbake, i must to run this command: bitbake openjdk-7-jre, Am i right?? and i think u exactly know this answer ^_^ Mar 11 08:59:50 LinuxMice: i actually fail to understand the question, and what makes you think that i specifically know it? Mar 11 09:00:07 LinuxMice: i guess you *want* to build that thing, right? Mar 11 09:01:52 LetoThe2nd: Yes i want to build java with bitbake command and this command bitbake openjdk-7-jre is right command ? Mar 11 09:02:06 besides that, about everything i know about java in OE is this: http://git.yoctoproject.org/cgit/cgit.cgi/meta-java/about/ Mar 11 09:03:19 Hi all. Mar 11 09:03:23 and generally speaking bibaking a package will compile and package it - not magically move it somewhere, especially not to a target or filesystem Mar 11 09:03:41 DISTRO_FEATURES_DEFAULT features are not enabled, if i set DISTRO_FEATURES are they? Mar 11 09:10:09 Anticom: thats what docs suggest, yes. http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-DISTRO_FEATURES_DEFAULT Mar 11 10:07:45 hi ! Mar 11 10:30:17 ok, thx, one more question and I send my patch :P Mar 11 10:31:22 my patch modifies meta/recipes-core/initrdscripts/files/init-live.sh but not the recipe itself... I would have expected to have to increment something... but there is no PR in the recipe, and the recipe has no upstream Mar 11 10:31:56 should I create a new file, create a new .bb with a new version in its name, or should I just push and assume prserv will handle it correctly ? Mar 11 10:31:58 bitbake sees that you've modified the file and rebuilds as required Mar 11 10:32:28 rburton, yes it does, my question was more "do I need to increase the PV in my patch" Mar 11 10:32:35 no Mar 11 10:32:45 as bitbake sees a file modification and rebuilds as required Mar 11 10:32:51 ok, cool, i'm good to go then Mar 11 10:39:44 and sent... hopefully I got it right :) Mar 11 11:45:02 IS there a specific Cross compiler I need to be able to Compile on windows for my Yocto device? Mar 11 11:47:37 cart_man: i think the meta-mingw layer works on enabling the sdk generation for windows dev hosts, but.... there be dragons! Mar 11 11:48:03 LetoThe2nd: AWwww.... Mar 11 11:48:23 LetoThe2nd: Ok soo do you think coding in normal GCC / G++ on Linux will work best? Mar 11 11:48:55 cart_man: s/work best/probably be the only thing that works close to reliable/ Mar 11 11:51:37 LetoThe2nd: So I can just work on Ubuntu standard compiler not even need to use a crosscompiler ? Mar 11 11:51:45 cart_man: nope Mar 11 11:52:07 fuk soo need to use the meta-mingw... so where do I get this meta-mingw ? Mar 11 11:52:37 cart_man: given the correct arch, using your distros toolchain might work for very limited, statically linked use cases. Mar 11 11:53:00 cart_man: for anything else, use the sdk functionality provided by OE. its there for a reason. Mar 11 11:53:22 LetoThe2nd: Ok I am actually trying to get to the point where i can use exactly that which you are speaking of Mar 11 11:53:29 LetoThe2nd: But I am a bit lost Mar 11 11:54:00 I have my IMX6 board running Yocto at the moment AND its on the network soo I want to start building apps for it in C++ from my Desktop computer Mar 11 11:54:38 cart_man: the bottom line is: you have an OE based build process that generates your image. so then just use that very same build process to provide the fitting sdk. Mar 11 11:55:26 cart_man: for standard cases, this means only after doing "bitbake your-image", a "bitbake your-image -c populate_sdk" Mar 11 11:56:53 which will provide you with an installable sdk, that brings the toolchain matching your target system, as well as the available libraries and a setup script for your environment to use it. Mar 11 11:59:13 LetoThe2nd: Ok so how does it work? Does the code gets transfered to your device and compiled on it or do you need to set up a compiler on your Host PC? Mar 11 11:59:36 cart_man: ermmm.... Mar 11 11:59:55 12:56 < LetoThe2nd> which will provide you with an installable sdk, that brings the toolchain matching your target system, Mar 11 12:01:07 LetoThe2nd: Ok soo I install that then onto my Desktop? Ughh im confused Mar 11 12:01:47 cart_man: you install it onto your development host. which is not necessarily a desktop computer ;-) Mar 11 12:01:58 cart_man: have a good look at http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#cross-development-toolchain-generation Mar 11 12:04:00 cart_man: or generally anything in http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#application-development-workflow Mar 11 12:04:33 cart_man: the alternative is, try the populate_sdk comamnd and tinker around with it to see what it does and what you get out of it. Mar 11 12:06:07 or alternative #2: hire a consultant Mar 11 12:08:39 LetoThe2nd: No I am just confused as to how Targeted compilations work Mar 11 12:09:25 you probably mean "cross compilation" Mar 11 12:11:11 How is network configuration done properly, when using systemd? currently we simply install a file to /etc/network/interfaces in a init-ifupdown.bbappend recipe that seems kind of wonky ot me Mar 11 12:11:50 LetoThe2nd: Well how to compile onto my target.. cross probably ues Mar 11 12:11:51 yes` Mar 11 12:12:03 cart_man: *sigh* Mar 11 12:12:40 LetoThe2nd: : ( Mar 11 12:12:46 cart_man: you really have to get the terms "toolchain" "compilation" "architecture" "target" and "host" straight first. Mar 11 12:13:02 cart_man: otherwise you just will misinterpret about everything you see. Mar 11 12:13:46 think in the mega manaual there was a step by step tutorial on some hello world recipe iirc Mar 11 12:13:52 check that one out cart_man Mar 11 12:14:17 Anticom: generally the application development section in the yocto docs is pretty exhaustive. Mar 11 12:14:51 LetoThe2nd: is that concerning my recommendation to cart_man or concerning my question? Mar 11 12:15:12 cart_man: this just isn't as simple as clicking the compile button in your favorite uC-IDE with predefined projects, sorry. Mar 11 12:15:32 Anticom: concerning the recommendation, yes. Mar 11 12:16:08 Anticom: concerning your own one, the problem is just that there is no systemd way of network configuration. its up to whatever network management engine you choose to accompany it. Mar 11 12:16:59 so no delicious abstractions for that yet? Mar 11 12:17:01 meh Mar 11 12:17:42 LetoThe2nd: How are you doing interface configuration in your projects? Mar 11 12:17:42 Anticom: well there ist systemd-networkd, but we told you about that already yesterday - so i implied by you not asking specifically about it that you decided against it. Mar 11 12:17:58 Anticom: we have some custom stuff in place Mar 11 12:18:24 LetoThe2nd: Checked back with my PM and he told me i should investigate further Mar 11 12:19:03 LetoThe2nd: i saw, that there was /etc/systemd/network or something where you can place your config when using systemd Mar 11 12:19:04 Anticom: um, so you just mentally erased what you learned yesterday and ask the same questions again to "investigate further?" Mar 11 12:19:46 iirc there was no exhaustive answer to that question Mar 11 12:19:49 Anticom: no. plain wrong. there is a specific place under etc where you you can place your config when using systemd_networkd Mar 11 12:20:07 systemd by itself just does not do that. no matter how often you repeat it. Mar 11 12:20:59 So in the end it also comes to installing config files to rootfs? Mar 11 12:21:03 Anticom: the exhaustive answer was "pick one of the alternatives, then one can discuss on how to use it." Mar 11 12:21:04 that's what my question is more about Mar 11 12:21:54 Anticom: well as your configuration probably is meant to be persistent, it will always mean some file operation of one kind or the other. that holds true for networkd, connman, nm, everything i know. Mar 11 12:22:56 shouldn't you generally choose your solution by matching its features against your requests, instead of looking at the path where it wants his config files? Mar 11 12:24:33 LetoThe2nd: Well as i wrote yesterday already from what i know systemd is somewhat 100% compatible with sysvinit files. The idea we had was switching to systemd before we launch our product so we can port the things from sysvinit to systemd gradually when the product has already been launched Mar 11 12:24:55 Anticom: ... which i think is a really bad idea. Mar 11 12:25:09 why? Mar 11 12:25:28 Background is, that systemd should improve our boot times since it starts services in parralel instead of sequencially Mar 11 12:25:54 systemd is very, very invasive. and it is really not sysvinit compatible, at least not in terms of "i just install systemd, it makes everything magically faster while nothing breaks." Mar 11 12:26:40 LetoThe2nd: And now i'm curious what exactly did break and in turn how much efford it would take to fix it again Mar 11 12:27:36 LetoThe2nd: also why do they claim its compatible when it's really not? Mar 11 12:27:59 Anticom: its "enough compatible for the common desktop usecase." Mar 11 12:32:53 Anticom: i really know that for most desktop linux distribution users systemd has the effect/impression of "hey we switched the init system and now everything boots magically faster". but under the hood that is just not true. it is not the reality. Mar 11 12:33:28 LetoThe2nd: it doesn't have to be 'magically' faster yet Mar 11 12:33:45 LetoThe2nd: Ok so this SDK Tool that you where talking about.. where does this get installed .. My Question is DOES bitbake Generate anything that you must use to "Prepare" the development system (My Desktop NOT the ImX6 board) Mar 11 12:33:54 The idea is more about preparing our production image to enable making it faster int he future by providing the proper tools Mar 11 12:34:04 Currently it only has to 'not break' Mar 11 12:34:50 Anticom: if your image is more than trivial (it has more than one binary that has to be kicked off during boot), switching to systemd will break stuff and eat your dev time. Mar 11 12:35:27 LetoThe2nd: Binary of what sort? custom or something from the oe layers? Mar 11 12:35:38 i mean ntp, syslogd, etc. seemed to be running just fine Mar 11 12:35:51 cart_man: really, this is all neatly documented, and however complicated it is not black magic. i feel that trying to explain those basics again through irc is not really a valuable use of time. Mar 11 12:36:19 cart_man: my suggestion would be, find somebody who has already done that once and have him/her show it to you in real life. Mar 11 12:36:39 Anticom: because those already have seen the proper dev work needed for the transition. Mar 11 12:37:15 LetoThe2nd: well then we actually do have only a single service that is brought up by us Mar 11 12:37:17 LetoThe2nd: My problem is I am stuck not remotely close to anybody know even knows Linux : / ... I will try the docs Mar 11 12:38:13 Anticom: then write a proper systemd unit file and have it kicked off. sacrifice a chicken in the meanwhile and pray that nothing that you implicitly took for granted breaks. Mar 11 12:38:27 cart_man: 1st step: install linux, 2nd step: learn linux (/join #ubuntu etc.) 3rd step: get poky set up in such way you can at least build core-image-sato and run it in quemu, 4th step: come back here :) Mar 11 12:38:50 LetoThe2nd: Are goats instead of chicken alright too? Mar 11 12:39:06 Chicken are out currently :/ Mar 11 12:39:25 Anticom: it becomes really, really nasty if you have runtime monitoring of your service in terms of watchdogging it, restarting/rebooting Mar 11 12:39:51 Anticom: Lol well I at least got to installing my own Yocto build on my IMX6 so I am a bit clued up haha Mar 11 12:40:30 Anticom: i can only repeat my practical experience - if you are not in the very, very early stages in a project - do not ponder switching to systemd. Mar 11 12:40:50 LetoThe2nd: First of all i need to get my image to build again. Just installed lttng tools on my host and the image doesn't build anymore due to invalid kernel config ._. Mar 11 12:40:52 Anticom: usually you can gain a lot more of bootup spped by other means. Mar 11 12:42:02 LetoThe2nd: even if it's only a single executable that runs our system? I got about a month of time to get it working. Mar 11 12:42:13 Anticom: depends. Mar 11 12:42:52 Anticom: i cannot comment in a resonably sane way what will break for your specific application and usecase. Mar 11 12:43:08 Sure there's no standard answer to this Mar 11 12:43:11 just saw a *lot* breaking unexpected and differently fomr sysV here. Mar 11 12:44:00 But back to the networking stuff. Googling arround for systemd_networkd yielded https://wiki.archlinux.org/index.php/systemd-networkd#Wired_adapter_using_a_static_IP Mar 11 12:44:19 I don't get how stuff like connman relate to this Mar 11 12:44:40 Anticom: why should connman realte to a systemd-networkd specific configuration file? Mar 11 12:44:59 Anticom: well as your configuration probably is meant to be persistent, it will always mean some file operation of one kind or the other. that holds true for networkd, connman, nm, everything i know. Mar 11 12:46:21 I was asking, whether the way to go was installing a config to /etc/systemd/network/ but you declined didn't you? Mar 11 12:46:49 I know i'm probably giving you a lot of *sigh*'s but i honnestly don't get it then :/ Mar 11 12:46:55 Anticom: read again. Mar 11 12:47:19 Anticom: i said that every one of those solutions will include some config file. i did not say that they all use the same. Mar 11 12:47:46 LetoThe2nd: i saw, that there was /etc/systemd/network or something where you can place your config when using systemd \\ Anticom: no. plain wrong. there is a specific place under etc where you you can place your config when using systemd_networkd Mar 11 12:48:03 Anticom: read again, one more time. Mar 11 12:48:08 So basically installing files there is the way to go? Mar 11 12:48:23 Anticom: i pointed out that this is the place for *SYSTEMD_NETWORKD* Mar 11 12:48:30 Anticom: not for *SYSTEMD* Mar 11 12:48:53 is this a bb var or what are you refering to? Mar 11 12:49:29 Anticom: i am totally not referring to anything OE specific in the least. I am *only* pointing out how these things architecturally work. Mar 11 12:50:43 Arch wiki any many other websites state stuff like: "All configurations in this section are stored as foo.network in /etc/systemd/network." Mar 11 12:50:53 So what is wrong about my assumption? Mar 11 12:50:57 I just don't get it Mar 11 12:51:01 bottom line. if you have something that works, don't break it. if bootup speed is your concern, there's a lot more options you have besides systemd Mar 11 12:51:55 Anticom: they are just all referring to one specific network management software called systemd_networkd, which is part of the systemd source distribution. none of these places on the web applies to anything else. Mar 11 12:52:23 so yocto doesn't use systemd_networkd ? Mar 11 12:52:30 *pling* Mar 11 12:52:34 now you got it. Mar 11 12:52:50 shall i get you the backlog line where we told you that exact fact already yesterday? Mar 11 12:53:06 LetoThe2nd: no need, i belive you Mar 11 12:53:36 So for dummies: systemd_networkd is a networkmanager that talks systemd api? Mar 11 12:53:44 Anticom: yeah. Mar 11 12:54:01 and even in yocto i have multiple alternatives? Mar 11 12:54:08 like e.g. connman? Mar 11 12:54:16 Anticom: and again, be precise. its not "yocto that uses". its "the poky distribution that uses by default" Mar 11 12:54:26 Anticom: exactly. pick whatever you want Mar 11 12:54:46 Any recommendations? Mar 11 12:55:08 or a good resource comparing them so i can pick one myself? Mar 11 12:55:20 Anticom: recommendation "match your requirement against the provided feature sets" Mar 11 12:55:34 requirement: should work Mar 11 12:56:06 no special requirements Mar 11 12:56:08 afaik Mar 11 12:56:23 Anticom: that also holds true for an ip= line in the kernel bootargs. please don't be ridiculous. Mar 11 12:56:27 so maybe small footprint and fast would be nice Mar 11 12:56:59 LetoThe2nd: we're not building an IOT device. So network connectivity has no special requirements really Mar 11 12:57:09 didn't want to be offensive Mar 11 12:57:18 Anticom: i guess the requirement are just not qhat you think they are. Mar 11 12:57:53 Anticom: ask your user interface people how they are willing to provide configuration data, parse it, etc. Mar 11 12:58:51 no UI Mar 11 12:59:08 basically it's a logging device Mar 11 12:59:12 Anticom: ok, so they order the device from you with one single fixed network configuration? Mar 11 12:59:21 yep Mar 11 12:59:23 Anticom: then ask your manufacturing guys what they need. Mar 11 13:00:25 just another thought to ease porting from sysvinit. If systemd has a network manager what about sysvinit? is it integrated tight into sysvinit whereas systemd has it nice and modular or is there any for systemd that sysvinit also uses? Mar 11 13:01:01 the other way round is, just include connman or networkmanager (who will probably 'jsut work') for trivial cases, and wait until it either goes unnoticed or blows up. Mar 11 13:02:46 LetoThe2nd: and how do i instruct bitbake to use either of them? is this sufficient? http://www.yoctoproject.org/docs/2.0/mega-manual/mega-manual.html#ref-classes-systemd Mar 11 13:03:45 Hi Everyone, i build java with this command: bitbake openjdk-7-jre and when i want to install this package on my arm board , i take an error : Bus error, wherem am i wrong ? Mar 11 13:04:39 LinuxMice: got some more verbose error information? 'bus error' isn't much of information Mar 11 13:05:09 LinuxMice: sounds like the architecture you built for doesn't match the package.... maybe. Mar 11 13:05:16 Verbose of error is this: Bus error , no another thing Mar 11 13:06:04 Anticom: see http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-connectivity/networkmanager/networkmanager_1.0.10.bb Mar 11 13:06:11 Anticom: it comes complete with systemd unit support Mar 11 13:07:19 LetoThe2nd: I thing i take error Which is Bus error Because I have no Space for installing ? Mar 11 13:07:50 LinuxMice: that is something you can easily check, right? Mar 11 13:10:33 Yeap You are right this way, i boot on nand flash Now i boot on sd card :) Finally i have 1.6 gb available space :) Mar 11 13:16:06 Hi all. Has anyone ever seen wic fail with an error like "exec_cmd: install -m 0644 -D /etc /etc returned '1' instead of 0"?? Mar 11 13:19:38 (I realize it's not really Yocto-related directly, but wic seems to be a pretty common way of packaging Yocto, so I was hoping someone has seen that error) Mar 11 13:20:54 I don't see a way to force a "verbose" output or determine where it's logging information at either. But that error makes me wonder if the return value is simply saying the directory exists and shouldn't really be flagged as an "error" Mar 11 13:23:37 eengie, you might need to file a bug report, the wic devs tend not to hang out on irc Mar 11 13:25:03 eengie: what happens if you run that command yourself? Mar 11 13:28:23 CroftonIwork: Thanks (and thanks again for the last week of helping me out too -- made loads of progress). I may have just sniffed out the issue. I think it's all related to that first question about how to deploy and link files over onto the SD card which ended up being a multi-part approach involving the deploy class (to move the files over to that folder), a pkg_postinst function to create the links after the files were insta Mar 11 13:28:36 rburton: same error as I get when it's in my script. Mar 11 13:29:05 I think where I may have goofed is that I changed IMAGE_BOOT_FILES to simply say "etc" rather than specify explicitly all of the files... Mar 11 13:29:37 There doesn't seem to be a way to access that variable from recipes, unfortunately. Mar 11 13:30:09 hi Mar 11 13:30:16 Hello Mar 11 13:30:24 eengie: i mean, run the install command Mar 11 13:31:14 rburton, thanks again for the help with nettle. I have just seen that the fix has been merged in the latest version of branch fido. Mar 11 13:31:49 rburton: sorry, not enough coffee yet... It says omitting directory. Mar 11 13:32:04 eengie: there you go then :) Mar 11 13:32:58 rburton: yeah, I forgot it was one of the changes I made to that layer's conf file...trying to let the user simply deploy files into a couple of documented folders and have wic blindly pick up those whole directories and deploy them to the boot partition. Mar 11 13:38:45 Weird, according to the docs, the "glob pattern" should work. So having a layer.conf with IMAGE_BOOT_FILES += "etc/*" should gather up everything in the deploy/..../etc directory. Right? (http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-IMAGE_BOOT_FILES) Mar 11 13:39:29 maybe dumb question, but doesn't it have to be += " etc/*" ? Mar 11 13:40:40 Anticom: not according to that link. Then again that link is specifying with an equals, and I'm trying to append since my layer is tacking on extra things to the SD card. Mar 11 13:44:04 += adds whitespace Mar 11 13:44:59 Interesting... I did an IMAGE_BOOT_FILES += "etc/*;etc/" (equivalent to what the link shows above for the glob format) and got an error that it was malformed. Mar 11 13:46:04 The trouble I've been having with how to add things to that variable is that the most important one, the top-level Xilinx BSP layer has a weak 'set' for the kernel and u-boot, etc. Then the vendor's BSP sets it. But if I specify it, the layer order overwrites the vendor's spec even if I set my layer priority to lower than theirs. Mar 11 13:46:32 And touching the variable from a recipe does absolutely nothing. Mar 11 13:52:21 rburton: is there a function I could insert into that variable that would recursively generate a list of directories and files from a certain point? I think I've seen something like that in other recipes using things like bb.utils but I'm having trouble finding it. Mar 11 14:24:23 I added PACKAGE_CLASSES = "package_ipk" to my local.conf, yet bitbake meta-toolchain tries to build nativesdk-rpm (which fails). Any idea how to completely get rid of RPM? Mar 11 14:31:44 why does that fail? Mar 11 14:32:11 tmcguire__: rpm is used e.g. for debugedit tool even when building ipk packages Mar 11 14:32:20 rpm is often built because it has tools for handing debug info and other things.. Mar 11 14:32:25 what is the error? Mar 11 14:32:36 and what version of rpm are you using? Mar 11 14:32:48 I'm trying to get Yocto dizzy to work. Mar 11 14:32:57 wow, that is pretty old Mar 11 14:33:06 why not try a newer version of the YP? Mar 11 14:33:12 yeah, but I need it to compare with some actual real project Mar 11 14:33:23 Missing or unbuildable dependency chain was: ['meta-toolchain', 'nativesdk-packagegroup-sdk-host', 'nativesdk-smartpm', 'nativesdk-rpm'] Mar 11 14:33:30 Hmm, maybe I should just edit the packagegroup Mar 11 14:33:53 I don't believe dizzy supported nativesdk-rpm.. so I'm not sure why it would be in that list? Mar 11 14:33:58 FYI, error is "undefined reference to `db3Free'", but I'd rather exclude RPM than trying to fix that. Mar 11 14:34:01 are you mixing dizzy with newer recipes? Mar 11 14:34:25 any comparison you do will be invalid w/ RPM availabel in the system as it's used to process various debugging and dependency information Mar 11 14:34:38 I don't believe I'm mixing anything, unless I did something wrong. FWIW, the actual image built fine. Mar 11 14:34:46 db3Free comes from BerkleyDB Mar 11 14:35:44 well, rpm-native is used internally for dependency information, right? That actually builds fine. What doesn't build is nativesdk-rpm, is that really needed? Mar 11 14:36:02 * tmcguire__ tries to find the packagegroup thing now and look at it Mar 11 14:38:28 Ok, found, let's see in what interesting ways it will break when I remove nativesdk-smartpm from that package group... Mar 11 14:40:46 Btw, "bitbake -g -u depexp nativesdk-rpm" didn't show the dependency chain, only when i tried to blacklist rpm, I got the error message "dependency chain was: ['meta-toolchain', 'nativesdk-packagegroup-sdk-host', 'nativesdk-smartpm', 'nativesdk-rpm']" Mar 11 14:41:18 Is there another way to get the dependency chain, apart from blacklisting the package with "PNBLACKLIST"? Mar 11 14:45:15 Hi all, I'm trying to build a Yocto image with opencv installed. So I added meta-oe to have the recipe available, I added opencv to IMAGE_INSTALL, opencv is building fine but is not on the image tar.bz2. I don't understand why? Other packages are installing without any problem. Mar 11 14:45:36 I'm building core-image-base Mar 11 14:49:37 zeechs_: iirc opencv doesn't produce a package called "opencv" Mar 11 14:50:14 clearly i need to go and beat the opencv recipe with the work-properly stick Mar 11 14:50:56 zeechs_: assuming you have a fairly recent oe, "oe-pkgdata-util list-pkgs -p opencv" will list the packages that it actually created Mar 11 14:57:19 Hi Evertyone, Finally I installed open-jdk-7 on my device, and i installed tomcat7, JRE_HOME Path needed by is tomcat7, Where find JRE_HOME, Wehere am i installed java Mar 11 14:58:02 rburton, thanks, it indeed lists all the packages built, which I can find under the "tmp/deploy/rpm/" directory. But how come these packages are not on my target rootfs? Mar 11 14:58:31 LinuxMice: wow, are you sure you have enough RAM to run tomcat? :) Mar 11 14:59:24 rburton, is there a more general command to do the recipe => package matching ? something based on PACKAGES or is that the best way ? (and is there a technicall reason or is it a case of "nobody implemente it") Mar 11 14:59:26 rburton, do I have to add each package to the IMAGE_INSTALL variable ? Mar 11 14:59:31 mborzecki: Yeap i have 512 mb ram and i use yocto fido 1.8 its enough for tomcat7 ;_ Mar 11 14:59:44 boucman_work: oe-pkgdata-util Mar 11 15:00:31 boucman_work: that is the general command. reading the recipe is the alternative but the package list can often be determined at build time. oe-pkgdata-util looks at what the recipe actually built. Mar 11 15:00:35 Is anyone else seeing no documents listed at https://yoctoproject.org/documentation ? Mar 11 15:00:53 hrm Mar 11 15:01:02 zeechs_: you'd have to read the recipe, but unless there's a meta-package that depends on all the libraries then yes if you want all of opencv installed. Mar 11 15:01:12 patches welcome it sounds like a lot of people get angry at opencv Mar 11 15:01:16 ok Mar 11 15:01:53 kergoth: `If there are no results for this category, something is wrong. Please contact the site administrator via the Yocto bug tracker.` Mar 11 15:02:21 yeah, but i have a bunch of extensions installed that occasionally break sites, so i don't trust that it's not on my end :) Mar 11 15:02:38 also https://yoctoproject.org/documentation/active might be a bit outdated Mar 11 15:02:43 i.e. ublock origin, chameleon, referrer control Mar 11 15:02:54 so wanted to make sure it was a problem in general, not just here Mar 11 15:03:55 rburton, ok I get it. I now have a better understanding of how the IMAGE_INSTALL works. Thanks a lot! Mar 11 15:20:05 Is BB_DANGLINGAPPENDS_WARNONLY disabled by default? Manual doesn't tell it **directly** but it reads like it's disabled if you don't activate it in your local.conf http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-BB_DANGLINGAPPENDS_WARNONLY Mar 11 15:44:33 Hello, I was wondering does yocto have an API for setting file/dir timestaps for all files in final sysroot? Mar 11 15:50:13 This has been driving me crazy for a few daze. I'm trying to set a value into /etc/version in my image. Currently the "DATE+TIME" is in there, we want "1.3" in /etc/version. I've tried setting $DISTRO_VERSION, also grep'ed the OE tree + bitbake tree but cannot find anywhere which sets a value into /etc/version. Mar 11 15:51:30 rofl Mar 11 15:51:50 There is something in toolchain-scripts.bbclass which sets something like $versionfile? Mar 11 15:52:03 from #oe Mar 11 15:52:04 Crofton|work: if you're still looking /etc/version gets written in lib/oe/rootfs.py Mar 11 15:52:18 * T0mW looks Mar 11 15:52:49 there is also an /etc/timestamp Mar 11 15:53:02 joshuagl> Crofton|work: meta/classes/rootfs-postcommands.bbclass Mar 11 15:53:36 Crofton, Thanks! wonder why egrep doesn't like python? heh Mar 11 15:55:54 I started looking to append to something like do_populate_sysroot to change the file. Mar 11 15:56:08 then figured "enough, go ask" Mar 11 16:04:25 Crofton, YAY, you da' man! Mar 11 16:04:38 http://stackoverflow.com/questions/35927774/how-to-get-rootfs-build-timestamp-at-runtime Mar 11 16:04:38 * T0mW buys Crofton a pizza Mar 11 16:04:50 someone asked something similar on stackoverflow Mar 11 16:04:51 Anticom: yes, it's disabled by default. normally dangling appends are fatal Mar 11 16:05:33 Crofton, yeah, google the crap outa' this one, tons of hits but not answers. Maybe I need to retake remdial-google-101 ? Mar 11 16:07:07 * zeddii is locked and loaded for ELC Mar 11 16:07:20 * fray will be there as well Mar 11 16:07:26 * zeddii has good memories of the last time it was in San Diego Mar 11 16:07:36 or I should say "hazy memories" Mar 11 16:07:39 :) Mar 11 16:08:01 this is more inland (by the real convention center) then the last one which was next to the airport in the marina.. Mar 11 16:08:19 kergoth: thank's Mar 11 16:08:19 I've been to this area a lot at the San Diego Comic-Con... Mar 11 16:08:34 lots of stuff within walking distance Mar 11 16:09:37 Crofton|work. fray is convincing me to leave late enough to stay for the oedam. Mar 11 16:27:10 zeddii, aweso,e Mar 11 18:03:15 halstead: I think we agreed the maintainance window was now. When you're done could you start a build of master-next please? Mar 11 18:04:24 RP will do. Thank you. Mar 11 18:04:38 halstead: thanks! Mar 11 18:59:39 grr, i'm really not happy with the shallow implementation, but i'm also having trouble finding any better way to do it Mar 11 18:59:40 hmm Mar 11 19:05:06 * kergoth grumbles **** ENDING LOGGING AT Sat Mar 12 02:59:58 2016