**** BEGIN LOGGING AT Thu May 20 03:00:43 2021 May 20 03:33:04 is UPSTREAM_CHECK_URI actually being used at all? http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/lib/oe/recipeutils.py#n972 May 20 03:33:55 python3-scons needs to be checking https url, but even with that change it still claims 3.1.2 is the latest. No. 4.0.1 and 4.1.0 have been released since. May 20 03:34:42 curl -s https://scons.org/pages/download.html | pcregrep -o1 "(?P\d+(\.\d+)+)\.tar" May 20 03:37:58 s/it/devtool latest-version python3-scons May 20 03:40:52 moto-timo: It's used a lot... maybe the regex is wrong? May 20 03:41:29 well, I pasted in the regex check above and it is right May 20 03:42:08 from what I see in recipeutils, it is not using anything but SRC_URI May 20 03:42:23 but it has been a 12 hour day and I am tired. May 20 03:43:19 * moto-timo goes to vegetate May 20 04:31:30 Hey guys May 20 04:32:25 Is the packagegroup-gnome-desktop verified on hardknott May 20 04:32:47 I see blank screen when I run "gnome-session" May 20 04:33:06 I tried running mutter it seems to be working May 20 04:33:31 But this gnome-session fails and also it is hardware accelerated using gles2 May 20 05:35:45 i think i'm gonna leave freenode, i'm still reachable by email if people need to get in touch, and i'm on oftc and libera.chat for various reasons. May 20 05:35:51 have fun, stay safe, etc May 20 05:46:29 I’m being spammed by html DMs May 20 10:04:58 Hi May 20 10:05:31 i am thinking update my repos from warrior to dunfell May 20 10:05:31 what should i follow May 20 10:11:12 The migration guide in the docs would be a good first step May 20 10:30:51 can you share any refrence qschulz May 20 10:34:45 https://docs.yoctoproject.org/ref-manual/migration.html May 20 11:53:14 slightly off-topic. Krogoth to thud upgrade. 64b kernel, 32b userspace (most of it). do_ni_syscall: 312 callbacks suppressed and QThread[2872]: syscall 397 "traces" is what I have on the screen for one particular app (only exists in 32b on the system, at least, lib32- packages only are installed) May 20 11:53:35 seems like syscall 397 is statx on arm (does not exist on aarch64 though) May 20 11:54:57 kernel is heavily modified 4.9 with Android patches (vendors <3). In krogoth the headers were from 4.4 vanilla, in thud it's 4.18. We do not modify the linux-libc-headers. May 20 11:55:07 any idea/pointer :) ? May 20 11:58:10 qschulz: no way out, fork linux-libc-headers to your effective kernel May 20 11:59:13 I vividly remember one long time contributor saying it shouldn't matter, newer headers with older kernels should just work May 20 11:59:30 but I can try and see where it brings me May 20 12:06:25 qschulz: there are corner cases, when kernel to userspace ABI and APIs are the same, then yes it will work. But if you run into problems like yours then linux-linux-headers need to be updated to the effective kernel ones. May 20 12:07:09 but for syscalls, they are supposed to be never removed, only additions right? May 20 12:07:40 I mean, that's the promise of the kernel community AFAIR :D May 20 12:07:57 I'll test and let you know May 20 12:08:05 I hope we don't have to come to this May 20 12:09:40 yea syscalls is clear, but ioctl's, sysfs etc APIs are like wild, wild west May 20 12:11:28 and depending on how users see those and enable/disable features, then linux-libc-headers may need to match the actual kernel. I've optimized a bit and copied the uapi headers out of kernel tree to linux-libc-headers so that a kernel change doesn't require full userspace recompilation. May 20 12:14:16 and like the "<3 BSP vendors" comment, upstream does a good job at ABI compatibility while $VENDORS don't have a bleebing clue and break them even with minor updates... May 20 12:14:56 yeah but this is a problem if you have multiple machines to build which have different headers (we have different git repos for each customer so that's fine to us but smh anyway :) ) May 20 12:15:09 yeah, I know... damn vendors :D May 20 13:28:49 mcfrisk: reverting back to 4.4 linux-libc-headers made the problem go away /me sighs May 20 13:42:46 Hi all! There is this concept of keeping multiple recipes for a package, some pointing to individual tarball versions (or git tags), and others which point to the master branch tip. Does someone know a package which can serve as nice example? May 20 13:56:13 manuel1985: how do you want to pick between versions? May 20 13:57:01 I was under the impression this is `PREFERRED_VERSION_packagename` is for May 20 13:58:07 I'm building poky-tiny (for initramfs). I get "ERROR: Nothing PROVIDES 'virtual/kernel' ... linux-raspberrypi PROVIDES virtual/kernel but was skipped: Set PREFERRED_PROVIDER_virtual/kernel to linux-raspberrypi to enable it". Yet, local.conf says "PREFFERED_PROVIDER_virtual/kernel := "linux-raspberrypi"". Why can't the variable be seen? May 20 14:00:17 somehow... "linux-dummy PROVIDES virtual/kernel but was skipped: PREFERRED_PROVIDER_virtual/kernel set to linux-yocto-tiny, not linux-dummy"... it seems like it's evaluated to a different thing May 20 14:00:55 manuel1985: yup, just wanted to check if you were trying to install two versions at once May 20 14:01:10 also... PREFERRED_VERSION will be setable only in .conf files, just so we're clear May 20 14:01:56 luneff: bitbake -e virtual/kernel | grep "^PREFFERED_PROVIDER_virtual/kernel=" May 20 14:02:03 this will tell you what it's set to May 20 14:02:22 if it's incorrect, remove the grep and pipe to less or more or whatever May 20 14:02:45 look for the line starting with PREFFERED_PROVIDER_virtual/kernel, then read the lines "commented" just above to understand where it's set May 20 14:02:52 something's probably overriding it May 20 14:03:15 manuel1985: my-recipe_0.2.1.bb my-recipe_git.bb my-recipe_2.5.bb May 20 14:03:41 manuel1985: then you select the version with e.g. PREFERRED_PROVIDER_my-recipe = "0.2.1" May 20 14:05:50 qschulz, nothing: ERROR: Nothing PROVIDES 'virtual/kernel'. The error is gone if I edit poky-tiny.conf to PREFERRED_PROVIDER_virtual/kernel ?= "linux-yocto-tiny" instead of PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-tiny" May 20 14:06:13 manuel1985: usually you have a my-recipe.inc which has all the things in common between recipes May 20 14:06:32 apparently poky-tiny.conf is evaluated after local.conf and overrides the kernel setting to the wrong one (which gets discarded due to MACHINE incompatibility) May 20 14:06:44 does it have to be "=" in poky-tiny.conf? May 20 14:07:01 luneff: PREFERRED_PROVIDER_virtual/kernel is usually set in machine configuration files so I'm not too surprised May 20 14:07:33 qschulz: Thanks, that's exactly what I had in mind. I didn't yet completely grok how to write recipe pointing to a specific tag version vs. one which tracks the branch tip. Do you happen to know a package which can server as best practice demonstrator? May 20 14:08:46 manuel1985: branch tip => SRCREV="${AUTOREV}" PV="+git-${SRCPV}" (or something along those lines, seach for AUTOREV in the docs) May 20 14:08:49 For example I don't understand why to use SRCREV=AUTOREV. Why do I need SRCREV anyway? I'm missing the complete picture here May 20 14:09:06 manuel1985: the git fetcher needs to know which commit to use May 20 14:09:30 autorev just means, go and check in the upstream branch if there's something new, if not use the one you already fetched May 20 14:09:44 if yes, then fetch/update the branch and point to that commit May 20 14:10:03 there's a need for a mechanism to say to bitbake what to do May 20 14:10:14 it's done by using AUTOREV in SRCREV that's it May 20 14:12:14 ah, it's easy, I can just copy the distro.conf to my own file and edit the "=" line to be "?=" and it'll work May 20 14:12:21 Do I need AUTOREV for git tags? May 20 14:13:16 hey and hello... May 20 14:13:16 is there one that can develop a special HDMI driver fro yocto? May 20 14:14:07 manuel1985: no, you can use tag= directly May 20 14:14:31 luneff: it does not make much sense honestly, a kernel is machine specific, so there's little point to have it defined in a distro May 20 14:14:44 it's defined there because you need a default in case nobody sets it May 20 14:14:47 (I geuss) May 20 14:15:19 that "=" is overriding my setting in local.conf and makes the build unbuildable. apparently, noone ever compiled poky-tiny outside x86 world May 20 14:15:24 manuel1985: https://docs.yoctoproject.org/bitbake/bitbake-user-manual/bitbake-user-manual-fetching.html#git-fetcher-git May 20 14:15:38 luneff: the point being, you need to set this in the machine configuration file May 20 14:16:00 and if it fails there, then please send a patch to fix poky-tiny upstream :) May 20 14:17:32 it was set in meta-raspberrypi, but was overriden afterwards by poky-tiny. yes, I think that's a case for a patch May 20 14:18:14 https://github.com/agherzan/meta-raspberrypi/blob/master/conf/machine/include/rpi-default-providers.inc here. A "weak weak setting" May 20 14:22:14 luneff: yeah that's expected, since poky-tiny is probably parsed before the machine, PREFERRED_PROVIDER_virtual/kernel is already set (not weakly set) May 20 14:22:19 so it does nothing May 20 14:23:24 I'll write to the mailing list May 20 14:25:18 have you tested with PREFERRED_PROVIDER_virtual/kernel = "" in the machine configuration file? May 20 15:10:06 I have an issue with a kernel module not being autoprobed at boot. `depmod -a xxx` once booted + reboot, make it probe just fine. This command is in the postinst script so I don't understand really May 20 15:10:17 (out-of-tree driver, just to be clear) May 20 15:10:33 I have a second out-of-tree driver, this one autoprobes just fine May 20 15:11:05 Note /lib/modules/*/modules.dep contains the second oot module but not the first May 20 15:11:26 qschulz, haven't edited meta-raspberrypi stuff, probably makes little sense May 20 17:50:06 * qschulz sighs May 20 17:50:15 seems like a multilib issue :D May 20 18:00:22 Remember Bitcoin in 2008??? Pi is a new digital currency developed by Stanford PhDs, To claim your piece of pi goto https://minepi.com and use "ilkde" as your invitation code. Get your piece of the pi now ! May 20 18:10:40 Remember Bitcoin in 2008??? Pi is a new digital currency developed by Stanford PhDs, To claim your piece of pi goto https://minepi.com and use "ilkde" as your invitation code. Get your piece of the pi now ! May 20 18:13:26 who has /kickban privs ? May 20 18:16:55 ndec: halstead: RP: at least one of those people I guess zeddii ? May 20 18:19:45 zeddii, qschulz: removed... May 20 18:20:27 rude May 20 18:23:35 💪 May 20 18:23:40 There we go. :) May 20 18:23:51 thanks halstead :) May 20 18:24:09 I'm not sure if RP or I got it first. May 20 18:24:39 halstead: I added to the akick list in the end May 20 18:25:23 ya'll are so rude! :D May 20 18:25:44 and thanks! May 20 18:27:30 RP: while you're here :D May 20 18:27:45 zeddii: I do this so infrequently I never remember how May 20 18:27:48 * paulg blames the new freenode admins May 20 18:28:14 I discovered why depmod wasn't working at build time for one specific kernel module May 20 18:28:36 it turns out our dependency chain requested this kernel module to be built in its multilib variant May 20 18:29:11 I had a multilib/lib32/lib/modules/4.9.54/modules.dep which was never merged into rootfs/lib/modules/4.9.54/modules.dep in the image workdir May 20 18:29:54 so now I wanted to override RDEPENDS_${PN} only for lib32 variant May 20 18:29:57 qschulz: does your recipe inherit module? May 20 18:30:01 RP: yes May 20 18:30:09 paulg, lets start a conspiracy theory / rumour that this is how the freenode takeover is raising funds. May 20 18:30:43 qschulz: in theory there should be no multilib variants of kernel module recipes May 20 18:31:03 qschulz: see multilib_virtclass_handler and the module-base reference May 20 18:31:09 (cont.) but to no avail, I think the entries in RDEPENDS for lib32 variants are prepended by lib32- in bitbake, so I can't override it with virtclass-multilib-lib32 for example May 20 18:32:10 the plan being having always the 64b variant of the kernel module recipe even for lib32- recipes that depend on it May 20 18:33:13 zeddii, they are involved in putting nanobots in the vaccine as well, so I've heard. May 20 18:33:30 RP: but yes, it makes no sense to have lib32 kernel modules (the recipe provides more than just kernel modules, nightmarish recipe :) ) May 20 18:34:26 so I'll go with RDEPENDS_${PN} = '${@"" if d.getVar("MLPREFIX") else "something"}' and add the "something" to the image recipe directly but :/ May 20 18:35:21 qschulz: I'm puzzled as the idea is that any multilib variants would be skipped and there should only be one recipe which provides the variants May 20 18:35:40 qschulz: you're working around it but the question about why there are variants at all remains May 20 18:36:05 Sorry, wasn't clear May 20 18:36:46 image recipe: INSTALL += "lib32-recipe-b", recipe-b: RDEPENDS_${PN} += "recipe-c", recipe-c: kernel module (+ other stuff) May 20 18:38:05 ok, recipe-b is optee-client and recipe-c is proprietary optee kernel driver May 20 18:38:22 so to me it makes sense from Yocto PoV to have this RDEPENDS May 20 18:44:40 halstead: btw, the irc logs are not available on https://www.yoctoproject.org/irc/index.html for two weeks May 20 18:45:11 *phew* May 20 18:45:21 * zeddii goes back on good behaviour. May 20 18:47:30 is there a reason why someone would want to have an NFS shared SSTATE in a container, and then rsync the local changes back to the SSTATE? I'm hearing about this secondhand, but seems insane to me. May 20 18:47:47 Ya, I had a few things that I more or less just let go, so the seller wouldn't have to eat the costs cause it wasn't their fault.. pandemic and all May 20 18:47:53 (tickets for events) May 20 18:49:05 qschulz: I think I kind of see. I suspect the bahviour may be different between the packaging backends even :/ May 20 18:49:47 * zeddii notes fray is in the wrong channel again :D May 20 18:49:53 ya, I just noticed May 20 18:50:04 wrong window grabbed focus right when I started typing.. May 20 18:50:17 I'd really like to know why that happens sometimes May 20 18:50:21 I actually live in fear of this. May 20 18:50:34 ohh I've typed various credentials due to that before.. May 20 18:50:51 thats the problem with touch typing and then looking elsewhere May 20 18:51:05 so I'm training myself to really not type all the off colour things I think ;) May 20 18:56:12 jonmason: it sounds crazy to me too fwiw May 20 18:59:08 zeddii, i am going in the opposite direction May 20 18:59:35 :D May 20 19:14:57 Is “bitbake -c checkpkg ” still supposed to work? May 20 19:15:48 We used to have to INHERIT += “distrodata”… but not since at least warrior May 20 19:17:08 http://docs.yoctoproject.org/ref-manual/variables.html#term-UPSTREAM_CHECK_URI May 20 19:18:45 Hi, I am trying to build packgegroup-gnome-desktop May 20 19:19:03 I running into build errors during do_rootfs May 20 19:19:08 Install 1496 Packages May 20 19:19:09 Total size: 205 M May 20 19:19:09 Installed size: 672 M May 20 19:19:10 Downloading Packages: May 20 19:19:10 Running transaction check May 20 19:19:11 Transaction check succeeded. May 20 19:19:11 Running transaction test May 20 19:19:12 Error: Transaction test error: May 20 19:19:12   file /usr/share/polkit-1/rules.d conflicts between attempted installs of gnome-control-center-3.36.4-r0.2.cortexa72_cortexa53 and polkit-0.116-r0.2.cortexa72_cortexa53 May 20 19:19:13 ERROR: Logfile of failure stored in: /workspace2/yocto/yocto-2021.1-rel-1/build/tmp/work/zcu106_zynqmp-xilinx-linux/petalinux-image-minimal/1.0-r0/temp/log.do_rootfs.15723 May 20 19:19:13 ERROR: Task (/workspace2/yocto/yocto-2021.1-rel-1/sources/core/../meta-petalinux/recipes-core/images/petalinux-image-minimal.bb:do_rootfs) failed with exit code '1' May 20 19:19:37 Any pointers? May 20 19:20:15 moto-timo: we got rid of that didn't we? May 20 19:20:39 Mads: please don't paste multiple lines like that but use a pastebin May 20 19:21:05 Sure May 20 19:21:15 RP: I think so, but still in the manual and still distrodata.Distrodata.test_checkpkg May 20 19:21:46 moto-timo: the test was rewritten but not renamed. I suspect we failed to update the manual May 20 19:22:19 I’m trying to debug why upstream version check is wrong for python3-scons and going down a rabbit hole May 20 19:23:46 Need more debug lines in recipeutils I think… May 20 19:23:59 moto-timo: can recipetool help? kanavin might have pointers too May 20 19:24:38 RP: that’s the problem, it isn’t coming to the right conclusion May 20 19:26:06 kanavin: any insight why pyrhon3-scons version check is wrong? May 20 19:26:31 kanavin: curl -s https://scons.org/pages/download.html | pcregrep -o1 "(?P\d+(\.\d+)+)\.tar" May 20 19:27:18 (I have a local patch for http—>https, but it isn’t the problem May 20 19:32:43 @RP: https://pastebin.com/DsxQRUyr May 20 19:34:09 RP am trying to build packgegroup-gnome-desktop and I get build errors at do_rootfs May 20 19:37:47 Mads: two recipes are trying to install the same file(s). You either need to prevent one from building or resolve the conflict May 20 19:38:10 The error message says almost exactly that. May 20 19:46:20 w00t! FedEx just dropped off my 10th anniversary Yocto Project swag: May 20 19:46:20 https://shop.spreadshirt.com/the-yocto-project/yocto+project+10th+anniversary+official-A609e70c45454db222085bab3?productType=812&sellable=ZbzeOgwNyjSQaD17ODLp-812-7&appearance=39 May 20 19:57:30 I didn't even know the shop was alive! I recall some email that it was being considered, but didn't see a "go buy stuff" May 20 19:58:45 Thanks moto-timo May 20 20:05:26 I ordered two shirts May 20 20:06:32 looks like mine left Nevada this morning May 20 20:07:50 did I miss the announcement of it ? just curious if I wasn't at a meeting or wasn't on a mailing list. May 20 20:12:16 zeddii: HA! i knew it May 20 20:12:20 option_ignore=$(($RANDOM%2)) May 20 20:12:55 zeddii: hasn't been publicised a lot until some people vet the quality first May 20 20:13:06 zeddii: I think the plan was at the summit May 20 20:15:43 gotcha May 20 20:15:57 tlwoerner, :D May 20 20:16:01 I found it on twitter.. May 20 20:16:16 twitter, eh. I don't feel so bad then May 20 20:16:24 it was on the official YP twitter acct May 20 20:16:39 we need a better influencer than fray to flog it though. May 20 20:16:52 * zeddii fires up his instagram May 20 20:17:51 call the jester May 20 20:18:27 what's funny. is he literally posted 30 minutes ago on FB about being an instagram influencer .. and I hadn't seen the post until after I typed that. May 20 20:18:33 big data is getting impressive. May 20 20:18:42 must be the new freenode overlords, selling my data. May 20 20:19:19 sorry, I do NOT have an instragram account.. so I'm not allowed to be an 'influencer' May 20 20:19:35 * zeddii looks at paulg to back up the latest conspiracy theory May 20 20:19:55 no insta ? the horror. May 20 20:22:53 i must have missed something... what's this freenode story? May 20 20:28:43 ah: https://hackaday.com/2021/05/20/freenode-debacle-prompts-staff-exodus-new-network/ May 20 20:28:55 turns out, news.google is still a thing :-) May 20 20:31:56 zeddii: would you like to advertise on Twitter? May 20 20:44:13 jonmason, rburton: Any interest in a hung arm host build: https://autobuilder.yoctoproject.org/typhoon/#/builders/97/builds/2753 ? May 20 20:44:31 abelloni: we also have an ltp one on offer: https://autobuilder.yoctoproject.org/typhoon/#/builders/95/builds/1916 May 20 20:45:09 RP: I'll leave that one to rburton. I'm behind on my SWAT duties May 20 20:45:11 RP, put it on ebay and see if anyone bids on it May 20 20:45:15 haha I'm on the same page as zeddii I didnt know the shop was alive May 20 20:45:18 ordering now :) May 20 20:49:05 I follow the jester. I’m in the know May 20 20:52:32 https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/HXI7CNJ5MOKAOGCBFOBYPJVO2TS7S5MH/ May 20 20:53:03 ^switch to matrix and freenode debacle May 20 20:54:26 https://www.kline.sh May 20 20:55:02 special torture for zeddii May 20 20:55:09 I have no desired to use a differetn service.. Maybe a different IRC _server_ if necessary, but I've already got slack, discord and IRC.. I don't need 'matrix' May 20 20:55:20 yah. no thanks. May 20 20:55:36 ditto May 20 20:55:44 I'm not monitoring any more inputs May 20 20:55:58 the fact I have three is 2, too many May 20 20:56:14 I have to use slack/discord/irc for the communities I am in May 20 20:56:32 moto-timo: irccloud is awesome btw. Thanks for the suggestion May 20 20:56:36 slack - work (and LF conferences), discord gaming and twitch stuff.. IRC 'everything else' May 20 20:56:47 At least discord is really IRC 'like' under the hood May 20 20:57:04 I have to use teams at work... it's awful May 20 20:57:06 i figured out the problem with thekernel recipe failing do_populate_lic: S = "${WORKDIR}/git" May 20 20:57:06 JPEW: glad you like it. Game changer for me being aware May 20 20:57:16 Ya, I have heard NOTHING good about teams May 20 20:57:50 Teams is ok. We use it internally. Better than Skype for business and share point May 20 20:57:59 teams is ok. we use it all the time for work May 20 20:58:13 Teams is functional enough, but I can _never_ find anything. The interface is just so much less intutive than everything else May 20 20:58:13 teams is ok.. if you compare to Skype for Business.. lol May 20 20:58:23 ms even provides a linux version. woo-hoo! May 20 20:58:31 I wish there was a way to disable Slack threads May 20 20:58:36 it would be far more useful then May 20 20:58:40 Given the approved tools, teams is like candy May 20 20:59:36 I've seen the same kinds of 'tools' for Slack.. (even developed a couple for discord) May 20 20:59:54 really easy to create a bot for discord that will send approval requests, and then listen for a response May 20 20:59:54 * mcfrisk ordered a lot from https://shop.spreadshirt.net/the-yocto-project May 20 21:00:44 I will say though, if it were not for Teams, I would have no functional Outlook calendar on Linux, so that's nice. May 20 21:00:59 Anyone fancy digging out the CVE fix for https://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2021-3507 ? May 20 21:01:14 JPEW: the worse thing about teams is that it's too flexible. the number of chats and teams meetings channels has become almost unmanageable. May 20 21:01:15 JPEW I already have said, you want me to have a functional outlook calendar? Then you better supply me with a Mac. May 20 21:01:29 Granted, it means _every_ meeting I sechedule is a Teams meeting, but that's seen as a plus now :) May 20 21:01:38 * RP is aiming for single figure CVEs against master this weekend May 20 21:01:42 I refuse to use Outlook, the UI drives me INSANE May 20 21:01:50 Lookout! May 20 21:02:07 i use both lookout and thunderbird - best of both worlds.. May 20 21:02:08 When I was heavily involved with IronPython, I really like the Gitter/GitHub/TravisCI integration May 20 21:02:11 My dad, a windows user, we put outlook on when he upgraded from Win 7 to Win 10.. he couldn't figure it out.. May 20 21:02:28 thunderbird or web-outlook May 20 21:03:06 it's sad when web-outlook is more usable then ACTUAL outlook May 20 21:03:41 i meant web-outlook. there is no outlook for linux (AFAIK) May 20 21:04:02 I've heard of people using it with Wine, but that sounds painful (and I have no idea if it actually works) May 20 21:04:18 i have the great fortune to work for a group in which everyone uses linux - lubuntu 18.04 LTS May 20 21:04:37 yeah I can barely keep up with irc I dont want any others May 20 21:04:37 windows is nowhere to be found.. May 20 21:04:40 Give me a Mac too fray (luckily they do) May 20 21:05:05 I just bought a new UPS for my telco rack. the network management card appears to be dead... so I went to APC's tech support.. they wanted me to run windows tools.. Umm sorry no windows machines.. May 20 21:05:24 then they wanted me to connect my laptop to the serial port.. "Umm, when was the last time you bought a laptop with a serial port" May 20 21:05:26 fray: that sounds like a freaking nightmare. i never have been a proponent of wine. May 20 21:05:34 (I have USB serial adapters at my office, but not at home) May 20 21:05:41 if i need windows that bad, i run it in a vm May 20 21:05:43 said screw it, and it's going back to Amazon as defective May 20 21:05:57 I have... way too many USB-serial adaptors May 20 21:06:17 I have windows on my Mac via bootcamp.. but come on, I'm not rebooting cause their stupid hardware is broken and their only software is windows based May 20 21:06:24 yeah, these days "serial" means usb. May 20 21:06:41 Me too JPEW May 20 21:06:48 The UPS even has a USB port.. but of course it's USB-HID, not serial.. May 20 21:06:59 And some dead boards that weren’t 5V tolerant May 20 21:07:01 it's like these guys are living 20 years in the past May 20 21:07:44 add to that, how difficult is it to create a multi-mode USB port.. they could default to HID, then change the mode to serial and bingo console.. but no.. that would make sense and maybe cost 1 cent May 20 21:07:53 But try to do automated hardware testing without a serial port and rapidly lose sanity May 20 21:08:12 lol May 20 21:08:51 But these days the tty is so slow it constantly mucks with other processes May 20 21:09:20 Funny thing, the one at my office is only ever been used to setup IT equipment.. and not actual boards.. since hte boards either have their own USB serial ports... :P May 20 21:09:25 (or actual consoles) May 20 21:09:28 Ya, for our labgrid tests we use the serial to ensure we can establish ssh, then use use ssh after that May 20 21:10:12 Try to debug boot loader without it. I dare you. May 20 21:10:16 serial is always the first interface when bringing up a board. May 20 21:10:41 thank for for usb/serial dongles and minicom May 20 21:10:45 thank God May 20 21:11:03 RP: usual sources indicate no patch in upstream qemu yet, https://ubuntu.com/security/cve-2021-3507 https://security-tracker.debian.org/tracker/CVE-2021-3507 and nothing in https://github.com/qemu/qemu/commits/master pops right up. Is floppy emulation enabled? May 20 21:11:15 But most production hardware no longer sends bios/uefi out serial anymore, if you even have one May 20 21:11:28 JPEW: sure, ssh is always more convenient once you get networking setup. May 20 21:12:19 mcfrisk: thanks, I was just looking at that as well May 20 21:13:37 moto-timo: i've not seen a bios for embedded systems May 20 21:16:33 yates: https://wiki.radxa.com/RockpiX May 20 21:18:56 moto-timo, inherit pypi overrides the upstream check settings in the recipe with something else May 20 21:19:23 and that something else reports there is no new version May 20 21:19:48 kanavin: aha! Thank you May 20 21:20:08 moto-timo, check why pypi.org reports what it does May 20 21:20:19 mcfrisk: Hmm, right, I was thinking there was a patch for that, sorry :/ I will see if we do enable tha May 20 21:20:31 kanavin: weak signal from the past sneaks through my foggy brain and remembers another knob May 20 21:24:03 kanavin: did we ever work out what the internal liblzma default for memory usage was? May 20 21:24:33 RP: I was not involved, was never a problem on my side :-/ May 20 21:26:39 mcfrisk: answer is yes, its the core fdc block device which several arches use May 20 21:28:13 Hmmm. pypi bbclass sets weak UPSTREAM_CHECK_URI ?= but the recipe sets it before the inherit May 20 21:35:39 moto-timo, yes something doesn't compute here May 20 21:36:03 bitbake -e points to scons.org May 20 21:36:10 vmeson: The best URL defense is a good URL offense! May 20 21:38:19 kanavin: I know the http url is redirected to https… so I thought that was it… but no May 20 21:39:20 moto-timo: SRC_URI is looking odd. I think the actual tarball needs to come first, and no leading space. May 20 21:39:38 kanavin: I think I’ll try going back to the default pypi check too May 20 21:40:27 kanavin: thank you for the extra set of eyes. Off and in for three days and I’m frustrated May 20 21:56:56 * RP is going to struggle with stable and master patches :/ May 20 22:07:09 kanavin: you nailed it. The SRC_URI += fir the license before the inherit was the culprit. I owe you a very nice beer May 20 22:16:32 moto-timo, should be SRC_URI_append May 20 22:17:22 👍 May 20 22:22:24 kanavin: interesting, it’s the order. Must come after the inherit May 20 22:23:54 += is fine, people just forget that ordering is important May 20 22:37:21 RP: yep May 20 22:38:34 It still feels like UPSTREAM_CHECK_URI should be explicitly preferred in recipeutils, but if the recipe isn’t broken the bit bake fetcher works fine May 20 22:38:59 Funky corner case May 20 22:53:45 kanavin: running auh on it now to dog food the tools and be lazy May 20 22:53:52 Not^ **** ENDING LOGGING AT Fri May 21 02:59:56 2021