**** BEGIN LOGGING AT Fri May 19 03:00:02 2017 May 19 04:25:01 Hi, I have been using yocto have been succesufully able to build and configure, thanks to a very good yocto documentation. May 19 04:25:12 Currently I'm blocked by one issue, I want to remove some of the utils which I'm not using from the packagegroup-core-full-cmdline-sys-services && packagegroup-core-full-cmdline-utils, How do I achieve that May 19 04:25:19 http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-extended/packagegroups/packagegroup-core-full-cmdline.bb?h=master May 19 04:26:17 I tried using BBMASK and RDEPENDS_remove and able to achieve what I want, but is there a better approch? May 19 05:01:14 madav_: probably the easiest thing would be to create your own packagegroup based upon that one and use that instead May 19 05:01:50 madav_: you could even copy and modify it in your own layer and provided your layer has a higher priority, your version would be used in preference May 19 05:05:00 bluelightning: thanks a lot for the input, I will explore on creating my own packagegroup May 19 05:57:15 morning all May 19 06:22:45 hi, who knows how to fix " No rule to make target `/usr/bin/g-ir-scanner-wrapper', needed by `Gst-1.0.gir'" thanks May 19 06:22:58 am i mssing some thing in my local configuration ? May 19 07:17:53 Warning - I'm still here ! Again, I really apologize for apparently being a bit thick-skulled. tThere's just somehting about yocto That I'm finding incredibly confusing. Hopefully rburton's example will help. (https://github.com/rossburton/customdistro.git) May 19 07:22:16 Just to be clear - the purpose of providing bitbake alongside oe-core (meta, meta-selftest, meta-skeleton) is that oe-core is tightly coupled to the bitbake version. This implies no effort to maintain backwards compatibility (as is done in CMake for example, with cmake_minimum_required) Is that correct ? May 19 07:26:09 And, as far as the Yocto "reference" project, and layers that I create, I will face the same problem - namely, there's no built in mechanism for backward compatibility check, or feature check. My project will also be tightly coupled to the bitbake version I use - right ? May 19 07:30:43 Smitty_, see https://wiki.yoctoproject.org/wiki/Releases May 19 07:31:56 ant_work: Thanks, but how does that answer my question ? May 19 07:32:09 the changes between bitbake releases were sometimes big enough to break compatibility with older metadata May 19 07:32:46 We shuold rename that channel to #PrayForSmitty ( it's a joke ) May 19 07:33:02 Hah, thanks, the vote of confidence I need ! May 19 07:33:33 Look, you guys have been very helpful. I really appreciate it. I'm sure you are finding this unbearable May 19 07:33:46 Smitty_ : It's a joke, nothing bad May 19 07:33:55 I know May 19 07:34:47 Smitty_ : I imagine that is pretty hard with a ton of experience in other environment tool to start in a new environment... And the main problem i guess is you came with some presumptions that make you far from the yocto philosophy and don't understand what is the mecanism. May 19 07:36:27 Smitty_: Your layer releases should state which poky/oecore/bitbake release they match -- typically people name their branches with same names as poky May 19 07:42:05 So, it seems that once I run oe-init-buildenv (which then calls a few other scripts), any changes I make to the configuration of layers (IOW the top level bblayers.conf - maybe this is the target bblayers.conf ?) *within my source tree* - not the build dir - won't be picked up on subsequent builds, since the layer bblayers.conf is not re-processed - is that right ? May 19 07:42:42 Smitty_ : The conf in layer, is to tell which recipe " can be parse " May 19 07:42:49 because it's been copied into the build dir May 19 07:42:56 Smitty_ : And most of time the conf is quite always the same, its a conf that say EVERY recipe. May 19 07:44:10 Smitty_ : I'm not familiar with the structure, but i think bitbake check for the local conf in build dir, then it checks in the bblayers to know which layer can be parsed. And then he go to every layer and check in the conf of the layer to know which recipe could be parsed. May 19 07:44:25 Smitty_ : But as i'm newbie, i'm maybe wrong, or missing some of the mecanism May 19 07:45:14 If I add a layer (in my source tree) to the target after having run oe-init-buildenv, and I don't re-run oe-init-buildenv, and I just re-run bitbake my new layer won't get built - right ? May 19 07:45:33 Smitty_ : If you add a layer after May 19 07:45:45 Smitty_ : The only things is to add in bblayers.conf, the layer you added May 19 07:46:05 Smitty_ : So that bitbake knows which layers are needed for the build May 19 07:46:25 So, bitbake is not detecting all changes to source tree, like a make system ? May 19 07:46:42 Smitty_ : Bitbake only do what you tell it to do """"for me """" May 19 07:47:14 Smitty_ : And when you do embedded systems, you would like to specify what you want, and not having undesirable things because of automatic process... May 19 07:47:27 Smitty_: oe-init-buildenv only sets your shell environment and makes some checks May 19 07:47:40 No, it does a *lot* more than that ! May 19 07:47:49 Smitty_ : Imagine if you add a layer of python with more than 200 recipes, making your image big with lot of thing not needed.... What is the better ? Putting what you need or what you don't need ? May 19 07:48:58 it calls several scripts within oe-core which then call other scripts, etc. I didn't bother loking through them all, but, the oe-init-env actually parses important information about your project May 19 07:49:15 Smitty_ : Important information that you have tells May 19 07:49:30 Sure, but it's a one off ! May 19 07:49:53 Smitty_ : The simple things is May 19 07:49:57 So, if I then change something about my target, I need to run oe-init-env again, not just bitbake May 19 07:50:15 no May 19 07:50:25 Smitty_ : oe-init-env it's just a script May 19 07:50:42 Smitty_ : for setting some environmental values May 19 07:50:44 So, it's not just environment, it's processing my source. May 19 07:51:03 Smitty_ : oe create a build dir, and let you do "bitbake" May 19 07:51:13 Smitty_ : And oe it's NOTHING linked to your source tree May 19 07:51:18 Smitty_: oe-init-env only makes sure you have the configuration that's required to successfully run bitbake May 19 07:52:06 nothing prevents you from modifying that config later and bitbake will notice those changes May 19 07:52:20 Sorry, but, I can see that it definitely is - it processed information within my source, it is doing more than just setting environment variables May 19 07:52:42 Smitty_ : not at all May 19 07:53:01 Smitty_ : It doesn't know what layer you have May 19 07:53:41 Come on, I can see it in oe-setup-builddir. That script is looking at the contents of my source files and constructing files in build dir May 19 07:54:05 Smitty_ : Try it and you will see May 19 07:54:20 I have - surely you have too May 19 07:54:33 Smitty_ : oe-init-dev IT'S JUST AND ONLY JUST to make bitbake command working and doing a builddir May 19 07:54:46 sed -e "s|##OEROOT##|$OEROOT|g" \ -e "s|##COREBASE##|$OEROOT|g" \ $OECORELAYERCONF > "$BUILDDIR/conf/bblayers.conf" May 19 07:54:55 Smitty_ : If you take a computer, a terminal, you source once, and you never close the terminal, you won't NEVER need to source again. May 19 07:55:01 Smitty_: Regardless of what the script does, it's well-tested that you can add new layers and/or modify layers without running oe-init-build-env again May 19 07:56:10 explain your comments in reference to that sed statement May 19 07:56:31 Where is that sed statement? May 19 07:56:38 oe-init-builddir May 19 07:56:43 Smitty_: surely you can see from context that's just creating a configuration _if it does not exist_? May 19 07:57:39 Smitty_ : Can you repeat, and read , again and again, the name of the script? May 19 07:57:52 And surely *you* understand that when I start with nothing but source, I start by running oe-init-build-env May 19 07:57:57 Smitty_ : oe init build env May 19 07:58:17 Smitty_ : and it was what it is, and it was how most of us use it for what it is May 19 07:58:41 Smitty_: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/oe-init-build-env does not call a script called oe-init-builddir for me. May 19 07:59:17 sorry, oe-setup-builddir May 19 07:59:29 cat oe-init-build-env | grep "sed" give me no result May 19 08:00:02 oe-init-build-env sources oe-setup-builddir which calls sed May 19 08:00:12 Smitty_ : for oe setup build i have done the grep , and the only line using it is : TEMPLATECONF="$TEMPLATECONF" $OEROOT/scripts/oe-setup-builddir May 19 08:00:24 Smitty_ : And what's the name ? "TEMPLATECONF" May 19 08:00:32 Smitty_ : And poky is a "reference distribution " May 19 08:00:44 Smitty_ : So that you have a kind of " example conf " insidre your builddir May 19 08:00:50 Smitty_ : But it's only an example conf May 19 08:01:04 Smitty_ : So that you just only to take out the layer ytou don't need and takes the one you need May 19 08:01:25 Smitty_ : like the meta yocto bsp etc May 19 08:02:55 Well, clearly the script just creates a bblayers.conf with absolute paths to your current source tree May 19 08:03:10 If you move the source tree, you'd have to re-create bblayers, otherwise it's not necessary May 19 08:03:59 Smitty_ : You can simply sum up oe-init-build-env as a script that give environmental variables so that you can use bitbake and create you a build dir with the files you need for configuring your project May 19 08:04:53 oe-setup-build-dir is in oe-core. It's fundamental to the use (or at least one imagines it is) of oe-core. oe-core is fundamental to the use of the Yocto "reference build". If I start with the reference build, I don't expect to change fundamental behavior like that. May 19 08:05:03 Is there anyone who got Mariadb working from the standard recipe from the meta-oe layer? May 19 08:05:36 Smitty_ : yeah and you won't May 19 08:06:01 Smitty_ : The only things you need, it's configuring in bblayers.conf and local.conf for your project May 19 08:06:11 Smitty_ : Making a custom layer with your own recipes May 19 08:06:18 Smitty_ : and deploy it to your machine May 19 08:06:49 Smitty_ : ( there is much more I know, like adding the meta layer of the BSP of the board you will need for example ) May 19 08:08:48 So, for my project to build, I expect to have to use oe-core *as is*. oe-setup-builddir (part of oe-core) creates bblayers.conf *for me* from my source tree May 19 08:09:04 Smitty_ : I guess you should spend more time to check how to configure a distro, a machine, doing a layer and your own recipe instead of spending time of looking a script who it's made for a simple task which doesn't need to dig in it May 19 08:09:15 Smitty_ : no May 19 08:09:29 no ? May 19 08:09:47 Smitty_: you are free to provide your own configuration: in that case a default one will not be created May 19 08:09:49 Smitty_ : You can do a script if you want so that it configures automatically May 19 08:10:05 Smitty_ : So that you can provide it to other engineers May 19 08:10:27 How would my config get from the source tree to build-dir ? May 19 08:10:42 Smitty_ : By doing your own script if you want May 19 08:10:49 Smitty_ : But for me it's the final step May 19 08:11:03 That would require changing oe-core behavior May 19 08:11:18 Smitty_ : Maybe i'm not understanding what is your need May 19 08:12:13 Smitty_ : I personnaly use yocto to do a distro for the need we want it with only the things i want and put it on my machine May 19 08:12:19 Maybe I should have an idea of your background in software development - because this seems such a fundamental issue to me (as a very experienced software developer) that I can't understand how you don't see the problem May 19 08:12:36 Smitty_ : There is no problem May 19 08:12:43 Smitty_ : You are creating problem ahaha May 19 08:12:47 If you refuse to acknowledge it, I guess May 19 08:12:51 Smitty_ : Yocto it's not CMAKE May 19 08:12:54 Smitty_: I've got a feeling you aren't giving us all of the context for your issues -- you are not the only experienced developer in this room :) May 19 08:14:39 Smitty_ : But as a very experienced software developer, I expect from you, that in one day you have already an image with a recipe made by you and with your library May 19 08:16:24 Yeah, I used the term only because sometimes I feel like perhaps I'm not being taken seriously May 19 08:16:39 Smitty_ : " very experienced" May 19 08:16:53 OK, OK May 19 08:17:22 Smitty_ : So you are telling to people that use yocto for long times and some of here have contribute a lot in that project that you understand the things more than them ? And you see lot of things that are senseless? May 19 08:18:01 No, I don't understand yocto at all, but it is not from lack of experience in development May 19 08:18:08 that's the point May 19 08:18:13 Smitty_ : Maybe you're true, but I expect from a "very experienced" software developer to things of his own presumptions, and be more attentive to what the "yocto developper" tells. May 19 08:18:40 Smitty_ : If people that contributes in yocto project, that do the documentation, tells you what it is, trust them. May 19 08:19:09 It's not a question of trust - I am trying to understand how it works - how I am supposed to use such a system May 19 08:19:21 this is the issue, not trust May 19 08:19:31 Smitty_ : But you have different level of understanding May 19 08:19:35 Smitty_ : For example May 19 08:19:40 Smitty_ : when you want to build a house May 19 08:19:51 Smitty_ : are you more about using tool, and trying to understand how to build a house May 19 08:20:15 Smitty_ : or are you more about knowing how a tool is made ( with all the physics fundamentals ) May 19 08:20:23 If you try to use a hammer to saw a piece of wood, you're not going to build a house May 19 08:20:30 Smitty_ : Like you would like to do the calculation of forces in your screwdriver to understand how it works May 19 08:21:29 Smitty_ : And the way you are trying to understand, it's not how to do a distro for embedded devices, but you are trying to check how yocto is made, and as yocto project is made from open embedded systems, you expect in one month to know everything of a project that makes more than 15 years to do with lot of people that contributes? May 19 08:22:13 Smitty_ : So if you're taks is to build that house. Doesn't focus on anything else. May 19 08:22:14 I don't need to calculate the forces, but I need to know that a screwdriver is made for a specific type and size of screw before I use it May 19 08:22:24 Smitty_ : Most of people didn't need to dig into the oe-init-build-env May 19 08:22:44 Smitty_ : In my point of view you are trying to check the calculation of forces =) May 19 08:23:01 Smitty_ : And we are telling you that the screwdriver is made for this May 19 08:23:02 Smitty_: Here's an example of how we do it: We have https://pbot.rmdir.de/QnU25_vcQ5YXOHzDzsd6Tw as conf/templates/bblayers.conf.template and tell our developers to run TEMPLATECONF=conf/templates oe-init-build-env May 19 08:23:12 Smitty_ : And the only things that you tell us, is " noooo" May 19 08:23:15 Smitty_: That way we can be sure our developers are getting the bblayers.conf we expect them to have May 19 08:23:53 Smitty_ : 95% of the person that works in yocto doesn't know how oe-initbuild-dev works. May 19 08:24:03 Smitty_ : it's a total presumption the "percentage" May 19 08:24:44 Smitty_ : But they know they need a oe init build dev to do the things May 19 08:24:52 Smitty_ : as you know that you need a knife to cut a beef May 19 08:24:54 Well, excuse me, but I find that kind of statement quite frightening. I don't know how this chainsaw works, but I'm just gonna start using it May 19 08:25:15 Smitty_ : As i said, there is many level of understanding May 19 08:25:47 Smitty_ : trying to understand oe(-init-build-dev, is like fi you trying to understand why a chainsaw cut the wood May 19 08:26:03 Smitty_ : Knowing how to configure bblayers and local conf, is knowing how to use the chainsaw May 19 08:26:58 Smitty_ : Do you expect from people working in building house to be very good in fundamentals physics? It's the image you give to me by digging into things that are unnecessary for building a distro for embedded systems May 19 08:27:05 So, I'm trying to build rburton's awesome example, and I run into this error: ERROR: linux-yocto-4.10.15+gitAUTOINC+4d929fac34_d2c1ed3c0c-r0 do_fetch: Fetcher failure: (etc...) May 19 08:27:15 Smitty_ : But i repeat, i'm also a newbie, it's my understanding of yocto. May 19 08:27:26 Smitty_ : And I don't have your experience, I might be wrong. May 19 08:27:35 Smitty_: the (etc...) part that you cut is the interesting part May 19 08:27:48 Smitty_ : So it's more interesting to dig into how to use bitake -e May 19 08:27:56 Smitty_ : to identify the problems May 19 08:28:02 ERROR: linux-yocto-4.10.15+gitAUTOINC+4d929fac34_d2c1ed3c0c-r0 do_fetch: Fetcher failure for URL: 'git://git.yoctoproject.org/linux-yocto-4.10.git;name=machine;branch=standard/base;'. Unable to fetch URL from any source May 19 08:28:16 Maybe my problem from yesterday with pgit proxy setup ? May 19 08:28:30 Smitty_ : try to do yourself directly from commandline May 19 08:28:42 try to check in internet also if that works May 19 08:28:47 Smitty_: can you pastebin tmp/work/*/linux-yocto*/*/temp/log.do_fetch? May 19 08:28:54 from where ? May 19 08:29:13 Smitty_ : usually, when you have error, there is a log with the full path to where it is May 19 08:29:13 It should have the detailed log output; Yocto re-tries from mirrors on error, which may hide the actual problem with the first mirror May 19 08:29:17 Smitty_ : pastebin it May 19 08:30:01 Can someone help me with this error? https://pastebin.com/ah4wmGVD May 19 08:31:12 rj_ : have you wrote that line? May 19 08:31:31 rj_: the function is there in current master: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/bitbake/lib/bb/utils.py#n991 May 19 08:31:46 rj_: maybe it's a mismatch of your layer version and yocto version you use? May 19 08:31:58 https://pastebin.com/cXiwEeVp May 19 08:32:14 neverpanic: How would i fic that? I already used git checkout morty on pretty much everything at this point May 19 08:32:16 bb.utils.filter() is pretty recent: http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/bitbake/lib/bb/utils.py?id=3004f57a25514926a0c688d333ea6e6ae3486acc May 19 08:32:55 rj_: did you grep for bb.utils.filter in your layers iand identify the recipe that filails? May 19 08:33:23 neverpanic: Mariadb fails May 19 08:33:37 neverpanic: from the meta-oe layer in meta-openembedded May 19 08:33:40 Smitty_: git.yoctoproject.org[0: 140.211.169.56]: errno=Connection refused, seems like your proxy is the problem May 19 08:34:23 rj_: what branches are you on in your poky and meta-oe layers? May 19 08:34:43 neverpanic: Should be morty if im correct May 19 08:34:58 I see about 10 of these before the error: WARNING: gnu-config-native-20150728+gitAUTOINC+b576fa87c1-r0 do_fetch: Failed to fetch URL git://git.sv.gnu.org/config.git, attempting MIRRORS if available May 19 08:35:08 But, they apparently get resolved May 19 08:35:23 and they're git May 19 08:36:37 neverpanic: yea, both say: Master May 19 08:36:44 rj_: morty bitbake doesn't have bb.utils.filter() yet, so I assume your meta-openembedded is newer? May 19 08:36:48 neverpanic: and then *morty in green May 19 08:37:04 Smitty_: They probably succeed because they fall back to some mirror May 19 08:37:27 neverpanic: aha... So that means i either have to get a new version of poky, or an older version of meta-openembedded? May 19 08:37:29 a non-git mirror (using http instead ?) May 19 08:37:55 For example, or a git mirror using http May 19 08:39:55 magically working when run bitbake again. May 19 08:40:08 rj_: your branches need to match May 19 08:40:19 rj_: http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/mysql/mariadb.inc?h=morty#n51 is the morty branch and doesn't contain bb.utils.filter() May 19 08:40:41 rj_: http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/mysql/mariadb.inc?h=master#n53 is the version on master, which uses bb.utils.filter() May 19 08:40:53 neverpanic: i found the morty branch version of the recipe, which uses bb.utils.contains(), i guess that should work right? May 19 08:40:55 both your poky and your meta-openembedded repositories need to use the same branch May 19 08:42:08 neverpanic: i cloned the master version of meta-openembedded yeah, so that was my mistake. May 19 08:42:19 neverpanic: Thanks! May 19 08:48:08 OK, when I run oe-init-build-env, I don't pass my custom target as an argument, only the desired build directory, which oe-init sets up for me. oe-init (by way of oe-setup-builddir) always looks for a file named $BUILDDIR/conf/bblayers.conf. If it doesn't exist (like the first time someone else retrieves my project), it is created by processing $OEROOT/meta/conf/bblayers.conf.sample. To me that says that bblayers.conf.sample, *is May 19 08:49:23 yet somehow every yocto developer is not bothered by that - and is somehow ignoring that. I don't get it May 19 08:52:00 Smitty_: your message was cut off after "To me that says that bblayers.conf.sample, *is" May 19 08:52:15 To me that says that bblayers.conf.sample, *is not a sample* It is the config May 19 08:52:47 Yes and no. It's the config template, but the sed command you found will be run over it to generate the actual config May 19 08:53:21 As you might have seen from my example, your bblayers.conf.sample can be as simple as OEROOT = "#OEROOT#", include "${LAYERS_DIR}/conf/my-real-config" or something May 19 08:54:13 So while bblayers.conf is being generated, it does not actually have to contain any relevant configuration; you can just put that config elsewhere (and probably should, so you can change it without requiring users to re-run oe-init-build-env or adjust their bblayers.conf) May 19 08:56:06 Also, you should really instruct your users to set TEMPLATECONF to whatever templates you provide with your project May 19 08:57:27 I am assuming that the use of oe-setup-builddir is not really optional, since it's part of oe-core, and called by *every* example init-build-env I have seen. Since it's not optional, and I expect my developers to not change my code, or eo-core code, then $OEROOT/meta/conf/bblayers.conf.sample is the config May 19 08:58:00 $OEROOT/meta/conf/bblayers.conf.sample is not used if TEMPLATECONF is set May 19 08:58:15 And as I mentioned, you should really instruct your users to set that to a path under your control May 19 08:58:55 OK, now we're getting somewhere May 19 08:58:56 And again, it's not the config, because of the sed replacement you saw. It becomes the config after that replacement. May 19 08:59:30 The .sample extension is weird, .template would be better May 19 08:59:46 But you know, in CS there are two hard problems: cache invalidation, naming things and off-by-one errors. May 19 09:00:40 Smitty_ : Just a question. Why do you refuse doing what most of the people are telling to do? It's make quite one week that we tell you do work on bblayers.conf and local.conf. May 19 09:00:48 Smitty_ : is it only because of curiosity? May 19 09:00:53 Not at all. May 19 09:01:05 Smitty_ : Or it's a kind of " i don't like to use something that i don't really understand the full process " May 19 09:01:27 Smitty_ : In that case, maybe in 2037 you will have your distro builded into your machine ahah May 19 09:02:09 I don't really need every detail, but I need answers to things I don't understand, or find (in my mind) wrong May 19 09:02:27 Smitty_ : You can know the development process, but don't need to check for things to deep inside May 19 09:03:21 Let's not get into the meta discussion again, shall we? I believe that's been done to death by now. May 19 09:03:28 Smitty_ : If it is for curiosity, i can understand as a passionned person. If you do this just for salary or whatever, is a kind of waste of time. Ahah May 19 09:03:37 neverpanic : sorry. May 19 09:03:42 That's not how I work - really. I'm a do it right the first time, kind of guy. So, for example, instead of just add my library to a depends somewhere (which everyone admits is a kludge), I need to understand how to get it done right. May 19 09:04:02 Smitty_ : So why don't you start trying to understand it ? May 19 09:04:20 Well, that's exactly what I hope I am doing May 19 09:05:09 Smitty_ : Not really May 19 09:05:28 Smitty_ : Instead of knowing how to use the machine, you are trying to know how the machine works ( in deepth) May 19 09:05:59 Smitty_ : If you follow the quickstart guide and the development manual ( which is big ) is enough to do much of your work i guess. May 19 09:06:19 Smitty_ : The question you should have is " how to add a library to my final image " May 19 09:06:43 Actually, only to the SDK, since the library is static. May 19 09:07:09 Smitty_ : instead of polluting the main discussion, maybe we can continue in " private". May 19 09:11:04 neverpanic: Okay, so i change to the morty branch version of open-embedded, but now the recipe only does the functions like do_populate_sysroot and do_populate_setscene, but it doesn't compile anything... May 19 09:13:28 Hi all, someonr here knows if it's possible to combine multiple conf-notes.txt file in order to display all images available in a project ? (i mean with several images in several meta-*) May 19 09:13:58 Hi, I am new to xen development on yocto project. Was wondering if xen-arm can be emulated using the qemux86-64 machine? May 19 09:19:46 I get this error while trying to build xen-arm for qemu machine ------------ ExpansionError during parsing /home/pranay/nonqemu/build/../meta-virtualization/recipes-extended/xen/xen_git.bb: Failure expanding variable XEN_TARGET_ARCH, expression was ${@map_xen_arch(d.getVar('TARGET_ARCH', True), d)} which triggered exception TypeError: getVar() takes at least 3 arguments (2 given) May 19 09:25:07 pranay: I'm guessing the layers you've combined aren't compatible with each other and that one of them might be on the wrong branch May 19 09:40:01 RP, will look into it. thanks May 19 10:01:17 Hi. Just switched to pyro branches and I'm getting a sanity checker error looking for sdl-config. I have ASSUME_PROVIDED="libsdl-native" and sdl added to qemu's PACKAGECONFIG. sdl-config is definitely installed and on my path. May 19 10:02:51 Is PATH pruned somehow, or do I need to add it to a list of host apps or something? May 19 10:18:11 Is there an easy way to upgrade to the pyro branch from the morty branch without losing files i modified? May 19 10:19:49 rj_: If you've committed your changes you could try to rebase them onto pyro. May 19 10:28:03 Hello everyone, I have the following issue, I am getting this error `Layer 'my-layer' depends on layer 'meta-security', but this layer is not enabled in your configuration`. I have added in the `meta-my-layer/conf/layer.conf` the line `LAYERDEPENDS_my-layer += "meta-security"` and in the `conf/bblayers.conf` I have added the meta-security layer. Is there somewhere else that I need to add the meta-security? I am using Ubuntu 16.04. T May 19 10:30:42 btw, my gemu/sdl sanity failure seems to have been some transient glitch. I removed sdl from the qemu-native build - all OK. Then I added it back in - still all OK. Puzzled, but meh. May 19 10:47:02 Does anyone know where I can find a guide on how to set up custom recipes for installing a python package? May 19 10:47:15 Which is on PyPi May 19 10:48:23 maybe this will help? http://stackoverflow.com/questions/36545388/use-pip-install-command-in-recipes-yocto-project May 19 10:50:16 arkver: I looked into this, but bitbake cannot find the right source for the package, and I cant figure out how to point it in the right direction. May 19 10:51:11 christianbillp: ok. I haven't tried it myself so can't be of more help. May 19 10:54:21 christianbillp: have a look at meta-python, which has hundreds of recipes for packages for pypi (and maybe the one you want) May 19 10:55:46 rburton: I think I might just be a little lost in there. May 19 10:56:08 As I understand it, the .bb file has to be named according to (pypiname)_(version).bb? May 19 10:58:16 christianbillp: actually you can probably just use recipetool to make the start of a recipe, assuming that you've checked the layer index to verify nobody else has built the recipe already May 19 10:58:43 recipetool create May 19 10:59:36 stavros: I am just answering my onw question, so I tried some other combinations for the `LAYERDEPENDS_my-layer`, like `security-layer` and `security`. With the first one I had the same issue, but the second one seems to be working ok. May 19 11:00:50 rburton: Excellent tip May 19 11:07:31 rburton: I set up the recipe with the tool and added setuptools3 and package_name, but in the end I still get this error: python3-pylyrics-1.1.0-r0 do_fetch: Fetcher failure for URL: 'https://files.pythonhosted.org/packages/source/P/PyLyrics/PyLyrics-1.1.0.tar.gz'. Unable to fetch URL from any source. May 19 11:08:57 2017-05-19 12:08:53 ERROR 404: Not Found. May 19 11:09:03 the URL is wrong May 19 11:09:17 find out what the right url is and fix the recipe May 19 11:12:39 I agree, the URL seems to be incorrect. The package only has one file on pypi.python.org and I tried point to that file with SRC_URI, but it didn't make any difference May 19 11:15:21 christianbillp : what was the error message? May 19 11:15:44 christianbillp : I mean, when you did point to file with SRC_URI i mean. May 19 11:21:26 When I point directly to the file, I just get a 404 as well May 19 11:22:41 Actually changin the SRC_URI doesnt change where it looks for the file May 19 11:23:05 then it must be looking at a different recipe May 19 11:23:26 $ wget https://pypi.python.org/packages/9d/74/8e92322cb1425739849eb719c3b5d8ca05a6b1f027829fd56b4db7c4b7c7/PyLyrics-1.1.0.zip works May 19 11:24:32 There is only one recipe with this name, I think. I just created it an hour ago. But is the a way I can point bitbake to a specific recipe May 19 11:25:04 share your recipe? May 19 11:26:00 https://pastebin.com/ZzDDxjRp May 19 11:27:09 ah, inherit pypi overrides SRC_URI for you May 19 11:27:38 if you use that you need to set PYPI_PACKAGE May 19 11:27:44 or, don't inherit pypi May 19 11:28:03 if you use pypi then you don't need to set SRC_URI at all, assuming its magic to work out the right URL works May 19 11:29:16 404 with inherit pypi and package May 19 11:29:28 So perhaps it can't find the right URL May 19 11:29:42 yeah just remove pypi from inherit and use the src_url you put May 19 11:32:25 rburton: Seems like it worked, just need to test it. How does it know if its python3 or 2? May 19 11:33:07 pylyrics is py2 May 19 11:33:57 oh no it actualyl works on both (shocked face) May 19 11:34:04 inherit setuptools3 May 19 11:34:20 and it does the right thing May 19 11:34:31 (or if you want py2, then just inherit setuptools) May 19 11:35:22 rburton: Ah ok, just need to skip inheriting pypi and use files directly if possible May 19 11:35:33 well we should fix the pypi thing May 19 11:35:57 (33 May 19 11:36:11 Thanks for all the help. Should be possible for me to build python recipes May 19 11:40:26 fixed it May 19 11:40:38 the author only submitted a .zip to pypi but we default to tar.gz May 19 11:41:01 Do you want me to try with package and pypi? May 19 11:41:08 i can pastebin the recipe May 19 11:41:41 Ok, Ill take a look May 19 11:44:50 christianbillp: https://pastebin.com/d6jWaZUi May 19 11:45:18 feel free to ask why a line is how it is if you don't understand May 19 11:45:26 (its only 7 lines after all) May 19 11:47:55 rburton: The second line LIC_FILES_CHKSUM. May 19 11:48:02 Im not really sure how this works May 19 11:49:26 And also, is it correctly understood that the version information is in the .bb file name? May 19 11:49:46 we demand that every recipe checksums a license statement in the source so we can tell if the license may change over time, like if one day they relicensed to GPL3 May 19 11:50:15 this package annoyingly (badly) just puts license=MIT in the setup.py and nowhere else, so that's the one line in setup.py that is checksummed May 19 11:50:31 most packages will have a file called LICENSE or something with the actual terms in May 19 11:51:07 and yes if there is a version in the filename then that is the version in the recipe unless you override it with a PV assignment inside the recipe May 19 11:52:21 rburton: Ok so for this package (https://pypi.python.org/pypi/tensorflow) the should the filename be python3-tensorflow_1.2.0.bb or python3-tensorflor_1.2.0.rc0.bb? May 19 11:52:55 1.2.0rc0 May 19 11:53:07 and hope that they can count and the next release is 1.2.1 May 19 11:56:50 Are there any tools to manage an sstate cache (eg. merge stuff into one SSTATE_MIRRORS dir, clean out old stuff etc)? May 19 11:57:06 rburton: Hopefully. Thanks for the help May 19 11:57:26 arkver: sstate-cache-management has powerful logic, or just use find to remove old stuff. you can merge by just copying files. May 19 11:58:08 arkver: i just use find $SSTATE -type f -atime +60 -delete to prune my sstate now and again May 19 11:58:55 rbuton: thanks! May 19 11:59:23 rburton: thanks, with extra spelling fix. May 19 12:02:34 rburton: Do you know anything about developing Yocto on the Xilinx Zynq? May 19 12:02:42 nope May 19 12:03:06 Ah well, worth a shot :) May 19 12:28:53 rburton: Can I use a .whl file directly in a recipe? May 19 12:35:52 christianbillp: no idea what a .whl file is May 19 12:36:42 christianbillp: Maybe you can install it with pip? May 19 12:40:08 rburton: Sorry, it's a python related file. I am just trying to work out how to install python packages with recipes May 19 12:40:51 rj__: I am using setuptools in my recipe, I think it uses pip? May 19 12:42:34 christianbillp: setuptools uses "python setup.py install" as far as i know May 19 12:42:49 christianbillp: To use pip: inherit pypi May 19 12:45:37 rj__: Ah right, it just seems that it doesn't work for tensorflow May 19 12:45:50 Or rather. I am doing something wrong May 19 12:46:18 Do I need to use pypi3 for python3? May 19 12:47:02 christianbillp: i'm not sure, but i guess so, since it's the same for setuptools/setuptools3 May 19 12:49:52 rj__: It doesn't seem there is a .bbclass file for it, so perhaps pypi depends on setuptools to specify its version? May 19 12:55:06 christianbillp: hmm, do you have python-pip added to your image? i think that was needed aswell May 19 12:56:47 rj__: I got some help from rburton setting up simple recipes which use PyPi, and it works pretty much like "pip install" May 19 12:57:00 christianbillp: ah alright ^^ May 19 12:57:19 rj__: However, for tensorflow it doesn't seem to work, as it looks for a tar-gz file, which isn't available. It just has .whl files May 19 12:59:39 christianbillp: hmm, you can also unpack the .whl file and install the files on the correct place yourself right? May 19 12:59:56 christianbillp: if i remember correctly i did the same with a custom .whl file May 19 13:02:23 rj__: Ill try to look into that. May 19 13:06:24 rburton: please have a look at this: https://pastebin.com/QB8iMgHE May 19 13:10:32 It seems I'm still experiencing a problem with git and my proxy. I am still getting this: https://pastebin.com/cXiwEeVp May 19 13:11:16 Yesterday somebody was about to explain that I needed to change my git proxy config, which apparently is currently inadequate May 19 13:12:34 I already have this (and only this) in .gitconfig https://pastebin.com/p8QZRW5W May 19 13:14:21 So, do I need some entry for the "git" protocol ? as in [git] proxy = http://blah.blah:9999 May 19 13:15:08 because as far as I know, I only have http, https and socks May 19 13:15:20 as possible protocols May 19 13:46:27 anyone here has experience building flatpak with yocto? May 19 13:51:23 i'm trying out this repo: https://github.com/klihub/meta-flatpak May 19 13:51:45 but yocto fails to handle this line: https://github.com/klihub/meta-flatpak/blob/master/recipes-flatpak/flatpak-predefined-repos/flatpak-predefined-repos.bb#L35 May 19 13:52:55 and of course now that i asked, i see that error is very specific and helpful: https://paste.fedoraproject.org/paste/2axjRSZIfN3rvbBPHljVbV5M1UNdIGYhyRLivL9gydE=/ May 19 14:22:01 zeenix: sounds like mismatching branches maybe? May 19 14:22:26 jku: i guess it's just that it's not been ported to latest yocto May 19 14:24:30 holly cow May 19 14:24:39 maintainer is my good friend Krisztian May 19 14:24:57 :) I thought you might know him May 19 14:37:39 jku: isn't he your colleague? May 19 14:37:50 jku: i mean working at the same office? May 19 15:03:36 zeenix: yes May 19 15:06:22 Yocto only in 4th place? c'mon! :-) http://www.cnx-software.com/wp-content/uploads/2017/05/IoT-Linux-Distribution.png http://www.cnx-software.com/2017/05/19/top-programming-languages-operating-systems-for-the-internet-of-things/ May 19 15:26:07 sgw_: what is needed to get a bootable image with meta-intel's default WKS_FILE=meta-intel/scripts/lib/wic/canned-wks/systemd-bootdisk-uuid.wks ? May 19 15:27:30 Could it be that it assumes that the image recipe inherits certain classes? May 19 15:28:59 meta-intel sets EFI_PROVIDER, and only live-vm-common.bbclass seems to check that. May 19 15:30:00 And if I understand that class correctly, it's only getting inherited by default if IMAGE_FSTYPES contains "live". May 19 15:30:39 When I just use IMAGE_FSTYPES="wic", the resulting wic file is unusable due to missing EFI bootx64.efi - but I could be wrong. May 19 15:31:53 I'm seeing a chrpath error "new rpath too larg; maximum length 275" when building u-boot-mkimage_2017.01.bb from the pyro branch. The error seems to have been addressed https://patchwork.openembedded.org/patch/34653/. I'm not working out of /tmp, my project work dir is /home/rschaefe/workspace/yo May 19 15:32:09 Is there a known workaround? May 19 15:55:32 I'm trying to get TCF to work with remote debug. It's already installed in the base image I got from device vendor, but according to https://bugzilla.yoctoproject.org/show_bug.cgi?id=5069#c4 it looks like debug support is turned off by default in the standad recipe. May 19 15:55:33 Bug 5069: enhancement, Medium, Future, david.reyna, NEW , Eclipse TCF can support arm cpu debug now,but tcf-agent_git.bb can't do it this time. May 19 15:56:17 The indicated changes seem to be to the recipe file? Is there a way to change the installation in place through opkg or do I remake from source? May 19 15:57:26 no git in this image either and none in opkg servers for this device (iotdk.intel) as far as I can tell, is that normal? May 19 16:14:26 any qt5 experts? Can't get it to see fonts May 19 16:14:31 so angry May 19 16:42:14 how is there a guide how to build an iso image/usb image on the yocto site ... May 19 17:11:05 sgw_, mythi: core-image-minimal for MACHINE=intel-corei7-64 does not boot automatically under qemu with ovmf (enters shell, then one has to invoke bootx64.efi explicitly) while it works for refkit images - any idea what the relevant difference might be? May 19 17:18:16 pohly: hi, looking May 19 17:19:19 sgw_: note that I am working with a bastardized config (basically Poky + meta-intel + refkit PR #147). A first step would be to verify whether you can reproduce that. May 19 17:20:17 Crofton: What version of QT? May 19 17:21:13 pohly: ok, a build I just did with meta-intel did boot directly May 19 17:21:37 sgw_: the partition type of the core-image-minimal seems wrong: /dev/loop0p1 2048 49167 47120 23M Microsoft basic data May 19 17:21:57 For refkit it is "EFI System". May 19 17:22:52 Crofton: Try adding PACKAGECONFIG_FONTS_pn-qtbase = "fontconfig" to your distro and liberation-fonts to IMAGE_INSTALL. That fixed it when I was having problems May 19 17:24:01 sgw_: I'm still using canned-wks/systemd-bootdisk-uuid.wks, which was removed in meta-intel master. May 19 17:24:37 Looks like a bug in the meta-intel that refkit is currently using... let me check that. May 19 17:26:13 pohly: I am using pryo with the patches that clsulliv sent and the sda is EFI GPT May 19 17:26:57 sgw_: what is the partition type of sda1? May 19 17:27:06 /dev/sda1 1 13 97088+ ee EFI GPT May 19 17:27:41 Hmm, why does it have that? OE's systemd-bootdisk.wks doesn't seem that different from what I was using from meta-intel. May 19 17:28:28 I believe the only difference is that the one we're using now has ttyS0 as well May 19 17:28:29 "bitbake core-image-minimal -c populate_sdk" on pyro branch fails for u-boot-mkimage recipe with "chrpath command failed with exit code 7". Log Exerpt: "new rpath '$ORIGIN/ (...) too large; maximum length 275" Logfile: https://pastebin.com/fSW4YH0y May 19 17:30:18 I've seen this error on the mailing list. Is there a known way to get around this? The cause in the other mailing list was unintuitively a too short of path (like buidling in tmp). My path is /home/rschaefe/workspace/yo May 19 17:30:49 clsulliv, sgw_: can you double-check which WKS file you are using and what's in it for the system partition? May 19 17:31:18 In refkit, we have --part-type C12A7328-F81F-11D2-BA4B-00A0C93EC93B. May 19 17:33:59 sgw_: which additional patches from clsulliv are you using? May 19 17:34:17 on pyro branch May 19 17:34:34 pohly: I don't think there are any additional patches, just head of pyro May 19 17:35:35 Updating meta-intel hasn't helped. Time to check out Poky... May 19 17:51:57 for life of me, I am not able to get ptests into my image for non-qemu h/w May 19 17:52:19 how is there a guide how to build an iso image/usb image on the yocto site ... May 19 18:05:53 black_13: it depends on which h/w you are building for May 19 18:06:07 but yes you can generate iso types images May 19 18:06:11 x86 for now May 19 18:06:17 with yocto look at IMAGE_FSTYPES May 19 18:06:29 yeah i saw that is the type "iso" May 19 18:06:33 or "live image" May 19 18:06:52 yeah live is probably what you want if you want live usb boot media May 19 18:07:05 do you set the meta var in local.conf May 19 18:07:06 you will get a hdd image which you can dd into a usb May 19 18:07:12 yes you can May 19 18:07:23 but there is more than one place? May 19 18:09:52 generally machines themselves suggest image types and some append them too May 19 18:10:47 ok. May 19 18:10:56 i have the yocto book. May 19 18:11:49 "Embedded Linux Systems with the Yocto Project" May 19 18:12:20 is it discussed in the book? May 19 18:12:30 it might have changed a bit but primarily it should be May 19 18:12:47 i have gotten to chapter7 and the books pretty much up do date May 19 18:12:52 to date i mean May 19 18:13:51 khem: thanks for the help May 19 18:16:56 core-image-minimal is not really just the image is all the things that it will take to make the image May 19 18:17:41 that is all the cross toolchain and that will compile the kernel and affiliated parts like the clib busybox May 19 18:18:06 a long long time ago i did about 2 years work in embedded linux (2.4 kernel) May 19 18:18:37 i want to make a very small linux for investigatory purposes May 19 19:07:44 any experts (or not) on qt5 and fonts May 19 19:10:12 Crofton: did you see my previous comments? May 19 19:11:06 JPEWhacker, no, ben in and out of room and had to take laptop with me :( May 19 19:11:50 Crofton: Try adding PACKAGECONFIG_FONTS_pn-qtbase = "fontconfig" to your local.conf and liberation-fonts to IMAGE_INSTALL. That fixed it when I was having problems May 19 19:13:20 You can put the PACKAGECONFIG in your distro.conf long term, but for testing local.conf is fine May 19 19:17:51 clsulliv: is ${@bb.utils.contains("EFI_PROVIDER", "systemd-boot rmc-boot", "systemd-bootdisk.wks", "mkefidisk.wks", d)}" supposed to use systemd-bootdisk.wks when a) systemd-boot or rmc-boot are the EFI_PROVIDER, or b) when both are selected? May 19 19:18:10 I suspect you meant a) but bb.utils.contains implements b). May 19 19:18:16 interesting, I'll give it a shoit May 19 19:18:26 also, anyone want a cat: https://www.petrescue.com.au/listings/543388 May 19 19:23:39 Crofton: i know a bit about QT5 (a bit) May 19 19:25:08 trying JPEWhacker fixes May 19 19:25:10 urg May 19 19:25:18 have to save my config.tx ... May 19 19:28:32 Crofton what are you using qt5 for? May 19 19:29:10 Crofton: sir May 19 19:42:20 used by gqrx May 19 19:42:33 need working fonts config May 19 19:45:21 are you using qt5 for embedded? May 19 20:13:21 pohly: correct, I noticed and sent a patch for that a couple hours ago May 19 20:48:55 how do cause a particular bbclass recipe to run? May 19 20:55:22 black_13: a bbclass != recipe May 19 20:55:26 so what are you trying to say May 19 20:55:40 it is May 19 20:56:23 i am still hung up on making and iso of core-image-minimal May 19 20:56:42 IMAGE_FSTYPES=iso, bitbake core-image-minimal, remember to have a MACHINE which supports ISOs May 19 20:56:42 IMAGE_FSTYPES, as you were directed to yesterday May 19 20:56:44 an iso rather May 19 20:57:15 yes i was told that yesterday May 19 20:57:35 so what doesn't work? May 19 20:57:43 it would have to be something like virtualbox or an actual machine May 19 20:58:30 MACHINE=genericx86 works if you have poky, or a meta-intel BSP May 19 20:58:45 modified the local.conf to have have IMAGE_FSTHYPES="iso" May 19 20:59:12 and bitbake core-image-minimal and walked a way for a few hours May 19 20:59:34 the tmp/deploy did contain several built files but no iso May 19 21:00:50 pastebin the contents of tmp/deploy/images/[machine]/ May 19 21:00:58 sure May 19 21:01:35 * rburton hands over to kergoth as its 10pm and i need to pack for a trip tomorrow May 19 21:03:05 unfortunately generic x86 is still running May 19 21:05:09 what should I see May 19 21:09:29 run bitbake -e core-image-minimal|grep IMAGE_FSTYPES= May 19 21:09:58 and see if it's set to iso in that context, or if something else overrode your setting (distro and machine configs are loaded after local.conf, so if a machine does IMAGE_FSTYPES= rather than IMAGE_FSTYPES +=, you'd lose your setting) May 19 21:10:21 doing IMAGE_FSTYPES_append = " iso" rather than just setting it directly would sidestep that possible issue, but you should use bitbake -e to see if that's actually the case here May 19 21:13:29 once "core-image-minimal" some parts will not be rebuilt? that is if i dont change the architecture or host machine will previous items like the toolchain be untouched May 19 21:14:26 is "iso" actually a valid IMAGE_FSTYPE? I think you're actually looking for IMAGE_FSTYPE = "live" May 19 21:17:57 good point May 19 21:18:45 but as rburton said IMAGE_FSTYPES=" live" or IMAGE_FSTYPES="live" May 19 21:19:44 IMAGE_FSTYPES=live, bitbake core-image-minimal on the command line? May 19 21:25:43 the results in the ~/poky/build/tmp/deploy/images/genericx86$ are at http://codepad.org/ARucmpHp May 19 21:25:51 hey there is an iso May 19 21:25:54 doo dah! May 19 21:27:12 :) May 19 21:27:32 now lets see if it actually works May 19 21:27:46 clsulliv: i think iso and live are subtly different May 19 21:29:46 rburton: iso images are built by image-live.bbclass, but iso is apparently a valid IMAGE_FSTYPE by itself May 19 21:29:59 oh May 19 21:30:17 so mister parkranger sir which is ... baseball season or wabbit season May 19 21:32:05 There's this mysterious variable in image-live.bbclass without a manual page: IMAGE_TYPES_MASKED += "live hddimg iso" May 19 21:32:05 I suspect this is what lets you do IMAGE_FSTYPES="iso" May 19 21:34:59 I went snooping through the builds looking for keywords iso May 19 21:35:42 but i have to get my Mom from the airport and i do not pick her up in a timely fashion she will kill me May 19 21:35:55 halstead: seems that patchtest is now working as expected but I forgot to save the timestamp before the system start failing May 19 21:36:18 halstead: so it will start doing its job on new patches May 19 21:36:37 halstead: I will check next Monday and let you know the results May 19 21:37:17 lsandov, Should I try to guess a timestamp based on the logs? May 19 21:38:18 hey it boots! May 19 21:38:25 but i dont know what the root password is May 19 21:38:46 lsandov: the problem is that moving time to the past may result in some false positives May 19 21:38:52 halstead: ^ May 19 21:39:16 halstead: let me check May 19 21:46:20 halstead: lets leave like this. I need to figure out a way to week run a cron that test untested patches without running the can-be-merge-on-master check May 19 21:46:21 lsandov, The last successful POST to patchwork from patchtest was at 15/May/2017:09:01:41 +0000 May 19 21:46:43 halstead: I see May 19 21:46:53 lsandov, If that's helpful. May 19 21:47:31 halstead: it is May 19 22:29:47 black_13: if you didn't set debug-tweaks IMAGE_FEATURE then root has no password set (so you can't login!), if you did then it has the empty password. May 20 01:39:51 how do i create a recipe of a live image like black13s_image it uses a certain kernel config and creates an iso image and so on **** ENDING LOGGING AT Sat May 20 03:00:02 2017