**** BEGIN LOGGING AT Mon Apr 26 03:00:18 2021 Apr 26 07:55:57 good morning Apr 26 10:09:45 can anyone tell me if I'm supposed to resend a mail to a mailing list if I forgot to join the ML group beforehand? Apr 26 10:10:18 is it completely rejected or is it just put in a queue for the postmaster? Apr 26 10:10:57 yo dudX Apr 26 10:11:22 i think its "just" waiting for moderator approval, then. Apr 26 10:41:48 qschulz: what did the bounce say? I think it would depend on a few things such as whether you're a member of other lists on that server Apr 26 10:44:04 Undelivered mail from my mail provider, don't know enough of mail protocols to say if there's no way the mail is still in a queue anyway? Apr 26 10:45:11 but I guess I can assume it's not Apr 26 10:45:20 resending it then, worst case scenario, it'll be there twice :p Apr 26 10:45:51 RP: and it's a mail "for you" BTW, not sure you'll like it :| Apr 26 10:48:01 qschulz: I think you're a little confused about the history :) Apr 26 10:49:18 qschulz: CLICKBAIT! CLICKBAIT! Apr 26 10:50:28 RP: oh, I'm not too surprised :D Apr 26 10:52:32 LetoThe2nd: https://media.giphy.com/media/B0vFTrb0ZGDf2/giphy.gif Apr 26 10:52:40 Trying to play with patches in kernel-meta definitions, I realize that I don't get any of those applied, they end up in .kernel-meta/unused.patch.queue, and no there is no .kernel-meta/patch.queue. I'm probably overlooking something, but I don't see anything in kernel-dev doc that would explain this. Any hints ? Apr 26 11:25:35 RP: thanks for the answer, but well... now the description of my talk is kinda bad :p Should have asked before yesterday... Me and procrastination, a long love story Apr 26 11:27:12 qschulz: well, yes. I suspect they might let you tweak it ;-) Apr 26 11:27:56 qschulz: don't get me wrong, removing overrides would make my life so much easier :) Apr 26 11:28:11 it would also break one of OE's most powerful features Apr 26 11:30:17 RP: yup, I agree, we make extensive use of it (probably too much :p) and the whole linux-yocto recipe is anyway a testament of how useful it is :) Apr 26 11:34:38 * LetoThe2nd really has to read all the proposals now. Apr 26 11:36:04 but i think its fair game to say that there is a lot of really, really good stuff! Apr 26 11:38:20 LetoThe2nd: good luck, have fun! Apr 26 11:39:28 will do Apr 26 11:40:52 qschulz: I wasn't trying to be negative about the idea of changing things, I just want to ensure the history is straight! :) Apr 26 11:44:15 HI Apr 26 11:44:40 can i PACKAGECONFIG in local.conf or shoukd it  be bbappend file Apr 26 11:44:47 use* Apr 26 11:45:16 Hi everyone! I've got a recipe "mender-connect.bb" which has "FILES_${PN}_append-mender-connect-systemd += "some file I'm interested"". How can I install that package? I assumed it would be "mender-connect-mender-connect-systemd" but bitbake tells me nothing RPROVIDES that. Apr 26 11:47:53 RP: I understood you but thanks for clarifying anyway :) Apr 26 11:48:13 Nithesh: bbappend file. though technically you can do so from local.conf but it's very bad practice Apr 26 11:48:41 manuel1985: can you triple check that it's REALLY FILES_${PN}_append-mender-connect-systemd the name of the variable? Apr 26 11:48:58 qschulz: I sometimes stand accused of not being open to new ideas and too negative, it is making me a bit paranoid! Apr 26 11:49:41 RP: isn't that the burdden that saddly most open-source SW maintainers have to bear? Apr 26 11:50:28 qschulz: probably :) Apr 26 11:51:12 qschulz: Yes it is. It really even has _append as well as +=. Apr 26 11:51:21 manuel1985: it is incorrect then Apr 26 11:51:42 manuel1985: FILES_${PN}_append-mender-connect-systemd should probably be FILES_${PN}-mender-connect-systemd_append Apr 26 11:52:06 in which case the package to include is: ${PN}-mender-connect-system Apr 26 11:52:31 qschulz: Alright I see. Will give that a try! Apr 26 11:52:39 https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/python/python3/python3-manifest.json#n760 Apr 26 11:53:06 i want to remove unittest from these manifest usig PACKAGECONFIG Apr 26 11:53:31 how to use in my bbappend Apr 26 11:53:41 to remove unittest Apr 26 11:54:19 i got to understand PACKAGECONFIG, implementation wise i am confused Apr 26 11:54:57 Nithesh: PACKAGECONFIG is NOT the mechanism that make a package go into your image Apr 26 11:55:28 oh i want to remove the package Apr 26 11:55:57 Nithesh: if it is not the mechanism to add a package to your image, it is also not the way to remove a package from your image ;) Apr 26 11:55:59 can you suggest a suitable method inn yocto Apr 26 11:56:54 according to manual it says it will block Apr 26 11:57:39 Nithesh: link to the manual Apr 26 11:57:41 please Apr 26 11:57:58 https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-PACKAGECONFIG Apr 26 11:58:08 Nithesh: but I stand corrected, there's a field for rdepends in PACKAGECONFIG Apr 26 11:58:16 PACKAGECONFIG blocks are defined in recipes when you specify features and then arguments that define feature behaviors Apr 26 11:58:53 This variable provides a means of enabling or disabling features of a recipe on a per-recipe basis Apr 26 11:59:10 ignore first one Apr 26 11:59:51 Nithesh: yes, at build time... because it is a **recipe** and not a package. Honestly, I don't know to which package the runtime-deps-for-f1 applies Apr 26 12:00:18 anyway, pretty sure there's no need for you to change PACKAGECONFIG for python3-unittests Apr 26 12:00:35 you should jsut not include python3-modules or python3-doctests in your image Apr 26 12:00:54 ok can you suggest a good practice to do this Apr 26 12:01:27 my goal is actual to strip down python3 to optimize its footprint Apr 26 12:02:13 "you should jsut not include python3-modules or python3-doctests in your image" -> any variable configuration Apr 26 12:02:23 Nithesh: https://www.yoctoproject.org/irc/%23yocto.2021-04-19.log.html#t2021-04-19T10:39:26 Apr 26 12:03:14 just add python3-core instead of python3 when applicable too (in RDEPENDS or IMAGE_INSTALL), will bring only the bare minimal Apr 26 12:03:27 PACKAGECONFIG ? Apr 26 12:03:28 if you really want the smallest python footprint Apr 26 12:04:08 Nithesh: some other recipes having python3-unittests or whatever in their RDEPENDS or in PACKAGECONFIG runtime-deps-for-f1 Apr 26 12:06:29 should i remove those Apr 26 12:08:28 qschulz Apr 26 12:16:12 @qschu Apr 26 12:28:58 qschulz  i installed python3-core and checked Apr 26 12:29:10 as u said it reduces my footprint Apr 26 12:29:39 if i am adding more required packages Apr 26 12:29:47 it is increasing Apr 26 12:30:47 so can i remove modules modules from the package which is not required for me Apr 26 12:31:04 https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/python/python3/python3-manifest.json#n760 Apr 26 12:31:37 it is taking from this json as i gice build for a particulra package Apr 26 12:31:55 https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/python/python3/python3-manifest.json#n200 Apr 26 12:32:19 how to remove modules from these packages Apr 26 13:43:13 Has anyone experience with running SSH over a U(S)ART line? Apr 26 13:55:51 manuel1985: interesting though. Don't you need a internet layer underneeth though? So maybe something about running PPP? Apr 26 13:57:31 zeddii: looks like I'm confused about how scc is supposed to be called: I expected it to get fed all scc's from the BSP (esp. patches), but kernel-yocto.bbclass seems to give it only the contents of KERNEL_FEATURES - what did I miss ? Apr 26 13:58:36 yann: rfc4251 says "The transport layer will typically be Apr 26 13:58:36 run over a TCP/IP connection, but might also be used on top of any Apr 26 13:58:37 other reliable data stream." Apr 26 13:58:47 section 1 Apr 26 13:59:15 manuel1985: no clue here, sry Apr 26 13:59:40 oh sry that was malinus_, sry yann Apr 26 13:59:42 manuel1985: why would getty not be adequate anyway ? Apr 26 14:00:31 Security. I'm thinking about maintenance access to deployed devices. Apr 26 14:00:44 manuel1985: interesting, I assumed it needed TCP/IP. Tell me how it goes :) Apr 26 14:01:32 manuel1985: just because something is technically allowed by spec, it doesn't mean that there's an actual implementation for it. plus, serial is not exactly a definition of "reliable", hence you would always have no add some intermediary layer Apr 26 14:01:36 hm, I had no idea remote access via serial lines was still a thing nowadays Apr 26 14:01:59 manuel1985: long story short: run slip over the line, and then ssh on top. Apr 26 14:04:30 yann: wdym. remote access? Like over internet? I think serial is still either the primary or at least secondary way of debugging embedded linux systems? Apr 26 14:04:33 LetoThe2nd: Thanks! Apr 26 14:04:57 LetoThe2nd: what's the difference between slip and PPP again? Apr 26 14:05:46 malinus_: ok, then if you have physical access, what's the deal with security ? Apr 26 14:06:25 malinus_: IIRC slip is the new fancy ppp? Apr 26 14:06:53 (don't take this serious, and especially not as authorative) Apr 26 14:07:53 malinus_: you're unlikely to have your creds stolen when you plug your serial cable on your desk, and if you're connecting to a console server, it would be unwise not to have that one physically secured. So I don't see where you have a security problem Apr 26 14:09:17 LetoThe2nd: I always thought that PPP had replaced SLIP - and wikipedia seems to agree: " SLIP has largely been replaced by the Point-to-Point Protocol (PPP)" Apr 26 14:09:35 yann: i happily stand corrected! Apr 26 14:10:27 np, that's all greybeard tech anyway :) Apr 26 14:10:51 yann: oh yeah I agree 100%, I think I just misunderstood your claim. Apr 26 14:10:59 i do have an increasingly greying beard, in fact. Apr 26 14:11:22 no beard yet, and I still knew about PPP, how do you explain that, huh? Apr 26 14:12:13 time travel. Apr 26 14:14:45 eh :) Apr 26 14:19:26 pppd is quite often how your cell phone data connection works... it "connects" the CPU to the baseband processor. Apr 26 14:20:02 it's many layers of historic extensions on top of each other. Apr 26 14:20:31 rfs613: yeah that's where I know it from. In which case it's basically "TCP/IP + other stuff, over serial" Apr 26 14:23:49 one of my collegues gave a talk about this stuff recently: https://www.youtube.com/watch?v=NPeMqK_vFFc Apr 26 14:24:40 zeddii: that is, if I am using a BSP that pulls a .scc with patches (and I do see it handled in source/.kernel-meta/) it is not sufficient to get the patches applied. OTOH if I additionally I add the same .scc to KERNEL_FEATURES, then the patches do get applied Apr 26 14:26:14 zeddii: the kconf part of the same .scc does get used, though (and triggers a warning since the option definition is added by the patch, making the issue readily visible) Apr 26 14:45:22 How is useradd bbclass supposed to work in relation to the host contamination check? The temporary user added might conflict with the user on the host. So chown-ing using the useradd (useradd bbclass) configuration might easily get into false positives with the host contamination check. How do people deal with this? Apr 26 14:48:43 alephan: wouldn't it all be caught by pseudo? Apr 26 14:52:39 In what sense LetoThe2nd ? Apr 26 14:53:07 alephan: in the sense of pseudo taking care of the uid/gid and permissions stuff Apr 26 14:56:42 It does. But useradd creates users before install time too. So that the install rule can pick the up. Install happens under pseudo anyway - so that is a given but the sysroot is populated with temporary users. Apr 26 14:58:13 I don't see any check against the host UID/GID to avoid this. Apr 26 14:58:23 Unless there is some hidden magic I miss. Apr 26 15:37:06 Hi folks :-) Apr 26 15:42:21 ljh: hi :) Apr 26 15:44:32 Anyway - I think that the assumption here is that SYS_GID_MIN and SYS_GID_MAX would give you the flexibility to avoid the user on the host. So using --system is a workaround. Apr 26 16:13:25 I've got the following recipes available https://github.com/mendersoftware/meta-mender/tree/dunfell-v2021.01/meta-mender-core/recipes-mender/mender-client, but when I do `oe-pkgdata-util package-info mender-client` it will only show me the 2.4.1 one. How can I point bitbake to the 2.5 recipe? Apr 26 16:14:52 Oh because I've got 'PREFERRED_VERSION_pn-mender-client = "2.4.1"' in my local.conf Apr 26 16:14:57 Well, I'm an idiot. Apr 26 17:10:43 armpit: seeing | /mnt/b/yoe/master/build/tmp/work/cortexa72-yoe-linux-musl/gnutls/3.7.1-r0/recipe-sysroot-native/usr/bin/aarch64-yoe-linux-musl/aarch64-yoe-linux-musl-ld: /usr/lib/libseccomp.so: error adding symbols: file in wrong format Apr 26 17:10:47 :( Apr 26 17:30:11 Guest8605, I would recommend opening a defect Apr 26 17:31:47 I think it could be specific to clang too Apr 26 17:31:53 let me build it with gcc Apr 26 17:33:56 k Apr 26 18:10:33 armpit: happens with gcc too Apr 26 18:39:52 Guest8605, thanks Apr 26 19:49:34 Guest3539, khem sent a patch Apr 26 19:51:17 armpit: about what? libbpf? Apr 26 19:51:39 hehe, wrong Guest Apr 26 19:51:49 sorry Apr 26 19:52:09 khem is in disguise sometimes. Apr 26 19:52:19 ha. it was khem all along Apr 26 19:52:35 Guest8605 Apr 26 20:35:06 I'm having a couple of problems related to the python requirements (specifically the json-schema lib, apparently) when running docker-compose for zeus. Works alright on dunfell. Any ideias? Apr 26 21:16:05 Does anyone know anything about nios2 in meta-zephyr? Specifically, any magic to get it to compile Apr 26 21:52:54 armpit: yeah I use matrix bridge to talk to IRC servers and it wants you to identify after few days, its a pain but its ok, matrix is far more useful Apr 26 22:47:47 khem, that makes sense... You are the Matrix Apr 26 23:58:57 Is there any way to have an npm recipe build faster? I'm just editing my bbappend on node-red recipe and it takes almost 20 minutes every time **** ENDING LOGGING AT Tue Apr 27 02:59:56 2021