| josch | minute: if you don't want to wait another 4 hours, the Debian buildds already are done building the package and you can grab them from the incoming repo here: http://incoming.debian.org/debian-buildd/pool/main/r/reform-firedecor/ | 00:09 |
|---|---|---|
| - chomwitt (QUIT: Ping timeout: 246 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 00:09 | |
| minute | josch: oh great, i can try it on my rk3588 reform now | 00:10 |
| minute | josch: it works! | 00:12 |
| minute | josch: this fixed the issue, nice. the package is the same version though, right? | 00:12 |
| josch | same version as what other package? | 00:15 |
| josch | the debian revision is incremented by 1 | 00:15 |
| - mjw (QUIT: Ping timeout: 260 seconds) (~mjw@gnu.wildebeest.org) | 00:17 | |
| minute | josch: ah yeah, nevermind. it's now 23-4 and before it was 23-3... | 00:17 |
| vagrantc | reform-specific packages in debian ... shiny! | 00:24 |
| minute | yessss | 00:35 |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 01:23 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 01:30 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 01:31 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 01:33 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 01:41 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 01:44 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 01:47 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 01:49 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 01:57 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 01:59 | |
| - mtm (QUIT: Ping timeout: 268 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 02:02 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 02:02 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 02:05 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 02:11 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 02:14 | |
| - nsc (QUIT: Ping timeout: 256 seconds) (~nicolas@i5E86127A.versanet.de) | 03:30 | |
| + nsc (~nicolas@i5C74DDFC.versanet.de) | 03:31 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 04:10 | |
| - skipwich (QUIT: Quit: DISCONNECT) (~skipwich@user/skipwich) | 05:01 | |
| + skipwich (~skipwich@user/skipwich) | 05:02 | |
| + Christoph__ (~Christoph@p54bf6382.dip0.t-ipconnect.de) | 05:13 | |
| - Christoph_ (QUIT: Ping timeout: 256 seconds) (~Christoph@p4fe73e33.dip0.t-ipconnect.de) | 05:16 | |
| + chomwitt (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 05:25 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 05:32 | |
| - Christoph__ (QUIT: Remote host closed the connection) (~Christoph@p54bf6382.dip0.t-ipconnect.de) | 05:42 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 08:10 | |
| - bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 08:18 | |
| + jacobk (~quassel@64.189.201.150) | 08:46 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@64.189.201.150) | 09:40 | |
| + mjw (~mjw@gnu.wildebeest.org) | 09:47 | |
| Twodisbetter | when upgrade lpc tools and reform tools is it necessary to reboot? | 10:59 |
| Twodisbetter | *upgrading rather | 11:00 |
| josch | Twodisbetter: necessary for what? | 11:03 |
| Twodisbetter | to reboot | 11:03 |
| Twodisbetter | when you upgrade reform2 lpc and reform tools | 11:03 |
| Twodisbetter | do I need to reboot is the question | 11:03 |
| josch | it depends on what you want to achieve | 11:04 |
| josch | sometimes reform-tools will change something about how the system is booted | 11:04 |
| josch | if you want that to take affect, then yes you can reboot | 11:04 |
| josch | but you don't need to | 11:04 |
| Twodisbetter | Also when those reform tools upgrades it always changes the reform standby script. I always forget how to find the file to remove the keyboard backlight section, as I don't wnat them turned on when resuming | 11:04 |
| Twodisbetter | I don't forget how, just where that script is | 11:07 |
| josch | Twodisbetter: do you mean /usr/sbin/reform-standby? | 11:08 |
| josch | Twodisbetter: and what do you mean by "it always changes it"? The last change in that script was last year october. | 11:09 |
| Twodisbetter | yes, thank you! | 11:09 |
| Twodisbetter | WHen you upgrade tools it seems to overwrite any modifications i make to that file | 11:09 |
| Twodisbetter | All I am doing is commenting out keyboard backlight on in the resume section | 11:10 |
| Twodisbetter | but when I upgrade the file is replaced and I have to recomment that out | 11:11 |
| josch | yes, that is by design | 11:12 |
| Twodisbetter | ok, makes sense | 11:13 |
| josch | if you want your custom standby scripts you could either dpkg-divert the file | 11:13 |
| Twodisbetter | I'll just make a note on where the reform-standby file is )/usr/sbin) | 11:13 |
| Twodisbetter | no, I would rather not mess up things that are supposed to happen, because there are good reasons behind those actions | 11:14 |
| josch | or you could overwrite /lib/systemd/system/reform-sleep.service by putting a service of the same name into /etc and modifying that to call your custom sleep script | 11:14 |
| Twodisbetter | SO just remembering where the file is it good for me. | 11:14 |
| Twodisbetter | leaving it like this means that if there are changes to the script due to architectural or driver based changes that I always inherit those things. | 11:14 |
| Twodisbetter | For me the Reform being rock solid is the most important thing. | 11:14 |
| - kensanata (QUIT: Quit: Ping timeout (120 seconds)) (~alex@user/kensanata) | 12:16 | |
| + kensanata (~alex@user/kensanata) | 12:17 | |
| minute | ok, back in action | 12:37 |
| minute | last night i worked on clearing out my email backlog. very interesting @ systemd-growfs josch | 12:53 |
| - mjw (QUIT: Ping timeout: 256 seconds) (~mjw@gnu.wildebeest.org) | 12:55 | |
| * Guest2486 -> mjw | 13:10 | |
| - Aard (QUIT: Ping timeout: 264 seconds) (~bwachter@edna-edison.lart.info) | 13:22 | |
| + Aard (~bwachter@217.11.60.132) | 13:25 | |
| minute | josch: lets say i have a new pipeline stage build_qcacld in reform-debian-packages that makes reform-qcacld2*.deb artifacts. how do i get the reprepro stage to include them? i don't fully understand the code. it looks like it picks up stuff from a "changes" dir but i don't understand what that is. | 13:42 |
| josch | minute: you would add your job to the dependencies of reprepro and that job is supposed to put the .changes file from your build in the changes directory | 13:45 |
| josch | reprepo will pick it up from that dir | 13:45 |
| minute | ahaa, so i don't think that dpkg --build creates a changes file? | 13:46 |
| minute | i am currently building the packages like this: i have a tree of files containing DEBIAN dir and a control file. i run dpkg --build on it. done | 13:46 |
| minute | so i guess i need to look into how to make a changes file | 13:47 |
| minute | > dpkg-genchanges | 13:47 |
| josch | i think reprepo can also work with plain *.deb files but you then might need to adjust the reprepo job accordingly | 13:47 |
| minute | ok | 13:48 |
| minute | the man page mentions > includedeb codename .deb-filename | 13:49 |
| josch | then that's probably what you want | 13:50 |
| minute | so i will try sth like: for d in reform-qcacld2*.deb; do reprepro includedeb "$OURSUITE" "$d" | 13:50 |
| josch | yes, probably | 13:58 |
| josch | you also don't need a new pipeline stage for your job | 13:58 |
| josch | i'd put it into the "build" stage | 13:59 |
| josch | just declare the dependencies accordingly | 13:59 |
| minute | josch: sorry, stage was the wrong term | 14:01 |
| minute | i put it in the build stage, perhaps these are called jobs? | 14:02 |
| minute | build_qcacld: | 14:02 |
| minute | stage: build | 14:02 |
| - mtm (QUIT: Ping timeout: 260 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 14:02 | |
| josch | cool! :) | 14:02 |
| minute | ok, pushed my punch cards to the mainframe | 14:03 |
| minute | lets see how many trials this'll need to work :D | 14:03 |
| minute | https://source.mnt.re/reform/reform-debian-packages/-/pipelines/1978 | 14:04 |
| minute | i'll work on other stuff in the meantime. | 14:04 |
| josch | minute: though you are not running "reprepro includedeb" anywhere, are you? | 14:09 |
| josch | minute: also, you seem to expect the *.deb in a directory called "linux" but they are in the "changes" directory | 14:11 |
| minute | josch: i thought i added that to the reprepro script, will check in a minute | 14:37 |
| minute | josch: ah ok. when building locally those debs end up in the linux directory | 14:37 |
| minute | ah damn i didn't push my changes to reprepro.sh | 14:41 |
| minute | ok, 2nd try https://source.mnt.re/reform/reform-debian-packages/-/pipelines/1979 | 14:43 |
| josch | minute: the linux *.deb are in the linux directory, yes but they are not listed to be kept as artifacts | 14:46 |
| josch | it looks okay now -- lets see what breaks :) | 14:51 |
| minute | ok :D | 15:24 |
| minute | thanks for spotting those things! | 15:24 |
| minute | josch: i wonder what we could do to make the kernel build faster :3 i wonder what is the bottleneck | 15:35 |
| josch | minute: more cores? | 15:36 |
| minute | hmm we already have 48 | 15:36 |
| josch | the debian kernel is just huge because it builds everything | 15:36 |
| josch | distcc? | 15:36 |
| josch | the reform kernel build times are already very fast because i disabled lots of stuff | 15:37 |
| minute | hmmmmm somehow it just seems slower to me than it should... if my lowly i9-9900k can build it in a similar time, on disk | 15:37 |
| minute | but maybe it's just the reality | 15:37 |
| josch | but if it builds the same source, then your i9-9900k might have fewer but faster cores? | 15:37 |
| josch | the recent error might be a missing "apt-get update" | 15:42 |
| grimmware | could be io bound instead of cpu bound. How big is the actual build on-disk? If you've got enough memory the quickest way to test might be repeating the build but in tmpfs. | 15:46 |
| josch | grimmware: on the CI the whole build is done in a tmpfs | 15:47 |
| grimmware | 🔥 | 15:47 |
| josch | the machine has 47 G of memory :) | 15:47 |
| minute | oh nooo @ this failure... | 15:52 |
| minute | why oh why did i remove the apt update | 15:52 |
| josch | to feel a bit better, remember that i used to wait six hours for a kernel build imx8mq :D | 15:56 |
| [tj] | how long does the kernel build take? | 15:58 |
| josch | [tj]: 32 minutes | 16:01 |
| josch | and another 2 minutes to install build dependencies | 16:01 |
| hramrach | some rpm based distributions generate dependencies based on the kernel module symbol CRCs | 16:03 |
| hramrach | like it imports a symbol it depends on it, it exports a symbol it provides it | 16:03 |
| hramrach | somewhat inflates the amount of kernel package metadata | 16:04 |
| josch | the debian linux packaging is horribly complicated -- nothing i've ever seen before | 16:06 |
| jn | it's quite arcane, really | 16:07 |
| q66 | i've seen worse | 16:07 |
| q66 | but debian packaging is overall kinda antiquated and lacking in many aspects | 16:07 |
| josch | q66: yes, certainly many things would've been done differently if Debian was created in 2024 instead of 1996 | 16:09 |
| yankcrime | that's why fpm exists | 16:09 |
| q66 | josch: right | 16:10 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 16:10 | |
| hramrach | what is fpm? | 16:10 |
| josch | q66: unfortunately we are now stuck with 35k packages that are done that way :/ | 16:10 |
| q66 | yeah i don't mean it negatively towards debian, legacy is legacy | 16:10 |
| josch | but there are distros like nix which do a lot of things nicer | 16:10 |
| josch | q66: i agree with your point but i think the bigger problem in debian is the strong package ownership | 16:11 |
| q66 | i still prefer debian to any other big distro fwiw, especially anything rpm-based | 16:11 |
| q66 | a key point in it is that debian is not corporate-controlled | 16:12 |
| hramrach | the problem with rpm-based distros is .. rpm | 16:12 |
| josch | q66: well... ;) | 16:12 |
| q66 | it may not seem like much but it does mean the project cares more about good interop with the others | 16:13 |
| q66 | while the likes of redhat mostly care about their own interests | 16:13 |
| josch | it is a bit better than that yeah | 16:13 |
| hramrach | Debian is not corparoate-controlled but it's also not corporate-sponsored which is probably one of the reasons for lack of developer power to provide fixes - another might be some legacy of lack of organizational flexibility | 16:14 |
| josch | corporations still have a big influence on debian. Many DDs are paid for their work and naturally they can do much more in a day than somebody working on Debian in their free time. | 16:14 |
| minute | aha, it is probably because i don't " extends: .setup" | 16:15 |
| josch | and since usually things happen because people find time working on them... | 16:15 |
| josch | minute: uh nice find! | 16:15 |
| josch | minute: in other news, i can now create a sparse system image | 16:15 |
| josch | ls -lh reform-system-ls1028a.img | 16:15 |
| josch | -rw-r--r-- 1 josch josch 5.1G Apr 5 16:10 reform-system-ls1028a.img | 16:15 |
| josch | du -h reform-system-ls1028a.img | 16:15 |
| josch | 3.7Greform-system-ls1028a.img | 16:15 |
| q66 | i maintain an independent distro and even though we have nothing to do with debian we end up using some of their system components (e.g. initramfs-tools and a few others) and i feel a lot more comfortable dealing with these than equivalent corporate-written projects (e.g. from fedora) | 16:16 |
| josch | minute: together with bmaptool this allows only writing 3.7G to the sd-card instead of 5.1G | 16:17 |
| josch | i had to write my own tool using lseek(..., SEEK_HOLE) and copy_file_range() because I didn't find an existing tool that would work like dd but respect sparse-ness | 16:18 |
| minute | josch: wow | 16:20 |
| josch | in case anybody knows how to improve this: https://superuser.com/a/1838012/235286 | 16:22 |
| hramrach | hm, I cannot post on SO. Tried to register once aand it's nuts. Initially you can only post questions and answers, not comment. Tried to ask a question got downwoted or something, got some negative points, could not do anything anymore, done. | 16:25 |
| josch | minute: would this help you with flashing 700 pocket reforms? i can create a branch with this if you would like to try out whether bmaptool is able to speed-up your processes by a bit or if it adds more complexity than it saves time | 16:25 |
| minute | josch: not sure, maybe it would help! i just need the files and commands instead of doing dd from the img on usb stick to mmcblk2 | 16:28 |
| josch | okay, i'll prepare something so that you can see if you like it | 16:29 |
| josch | i never used it before myself, so i cannot tell you how easy/hard it is :D | 16:29 |
| josch | but according to the docs it should just be: bmap copy system.img /dev/mmcblk0 | 16:30 |
| josch | if nothing else, this exercise made me learn about the copy_file_range system call which seems to be relatively recent and according to the docs only stabilized in kernel 5.19 | 16:31 |
| minute | josch: sounds good @ bmap copy | 16:38 |
| + jacobk (~quassel@64.189.201.150) | 17:12 | |
| minute | > /usr/bin/bash: line 137: cd: changes: No such file or directory | 17:15 |
| minute | ok so so the linux artifacts don't seem to be there | 17:15 |
| minute | ls in pwd (/builds/reform/reform-debian-packages) yields absolutely nothign | 17:16 |
| minute | josch: gah i think i need another stage? because artifacts get passed only from stage to stage, not to other parallel jobs maybe? | 17:19 |
| josch | minute: the job should not be parallel if you have a depends relationship between them | 17:20 |
| minute | yeah it's not actually parallel. i mean they are in the same stage | 17:21 |
| minute | i have a dependency on build_linux | 17:21 |
| minute | but both jobs are in the build stage | 17:21 |
| josch | as far as i understood the gitlab docs, you don't need stages if you don't want to | 17:21 |
| josch | stages are a different way to think about dependencies | 17:21 |
| minute | josch: where can i see that? and how would i find the artifacts? | 17:22 |
| josch | unfortuntaly i have my daughter here now -- i can only really look at it again around after 22:00 | 17:22 |
| minute | ok thanks, then i have to figure it out myself now... probably i have to introduce another stage but we'll see | 17:22 |
| minute | > You can use the 'needs' keyword for this | 17:23 |
| minute | not sure if that really exists | 17:23 |
| - chomwitt (QUIT: Ping timeout: 255 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 17:24 | |
| minute | > Use artifacts: true (default) or artifacts: false to control when artifacts are downloaded in jobs that use needs. | 17:24 |
| minute | https://docs.gitlab.com/ee/ci/yaml/index.html#needs | 17:24 |
| minute | > You should not combine needs with dependencies in the same job. | 17:25 |
| minute | > Use the dependencies keyword to define a list of specific jobs to fetch artifacts from. The specified jobs must all be in earlier stages | 17:25 |
| minute | > must all be in earlier stages | 17:26 |
| josch | oooooh | 17:50 |
| josch | sorry, then i was wrong | 17:50 |
| josch | and your hunch was spot-on | 17:51 |
| ericsfraga | you can have file wide src block header arguments using #+property: lines | 17:55 |
| ericsfraga | (sorry, wrong room) | 17:56 |
| josch | minute: are you going to retry the reform-debian-package pipeline this evening? otherwise i'd squeeze a run of reform-system-image in | 18:28 |
| josch | i don't want to cut into your CI time :) | 18:29 |
| minute | josch: i will retry it until it works yeah | 18:37 |
| minute | argh, wrote cd.. instead of cd .. | 18:37 |
| minute | but at least the packages were identified correctly | 18:37 |
| minute | > LINUX_SRC_DEB=linux-source-6.7_6.7.9-reform2-2+reform20240405T152722Z_all.deb | 18:38 |
| minute | > LINUX_CFG_DEB=linux-config-6.7_6.7.9-reform2-2+reform20240405T152722Z_arm64.deb | 18:38 |
| - ericsfraga (QUIT: Quit: ERC 5.6-git (IRC client for GNU Emacs 30.0.50)) (~user@2.28.19.21) | 18:38 | |
| + ericsfraga (~user@2.28.19.21) | 18:40 | |
| minute | here we go again (5th try) | 18:40 |
| josch | fingers crossed! | 18:41 |
| josch | minute: didn't you want to create a new stage for build_qcacld so that it's able to fetch artifacts from build_linux? | 19:02 |
| minute | josch: no, i use needs: | 19:16 |
| minute | josch: needs can take artifacts from jobs in the same stage, unlike dependencies: | 19:18 |
| josch | huh interesting | 19:38 |
| minute | another blunder! 6th try going | 19:41 |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@64.189.201.150) | 19:45 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 19:46 | |
| + Christoph_ (~Christoph@p54bf6382.dip0.t-ipconnect.de) | 19:47 | |
| josch | every try only takes 50 minutes | 19:48 |
| josch | when i upload to debian, i have to wait 6 hours for things to change in the archive :D | 19:49 |
| minute | one needs to be patient :3 | 19:50 |
| josch | it's the difference between my hobby and your job on which your livelyhood depends -- waiting is much easier for me | 19:52 |
| minute | yeah, sometimes i'm stressed because every hour or day i spend here costs a lot of money (cost of living + cost of running the workspace etc) | 19:53 |
| + gustav28 (~gustav@c-fd68524e.019-141-67626730.bbcust.telenor.se) | 20:02 | |
| - xktr (QUIT: Ping timeout: 260 seconds) (~xktr@user/xktr) | 20:10 | |
| + xktr (~xktr@user/xktr) | 20:11 | |
| * mjw -> Guest3198 | 20:14 | |
| - Guest3198 (QUIT: Killed (silver.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:14 | |
| * mark_ -> mjw | 20:14 | |
| + Guest3198 (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:15 | |
| + jacobk (~quassel@utdpat242060.utdallas.edu) | 20:15 | |
| Twodisbetter | minute: no stress or anything, but do you have any news on the pocket and suspend? No worries either, no news is not bad news. | 20:30 |
| minute | Twodisbetter: last time i checked it worked, but audio didn't work after resume. i will soon check how it is working in the meantime with the current kernel | 20:44 |
| Twodisbetter | minute: fantastic! Awesome work as usual! | 20:45 |
| minute | maybe as a workaround we can reload audio related module | 20:45 |
| Twodisbetter | seems like it would probably work, I believe we are basically doing similar things on the reform itself. | 20:45 |
| josch | minute: about the latest error, your .deb seems to be missing the Section: field | 20:53 |
| josch | minute: you might want to add "Section: kernel" to https://source.mnt.re/reform/qcacld2/-/blob/binary-deb/debian/DEBIAN/control?ref_type=heads | 20:55 |
| minute | josch: oh i just wanted to ask | 20:58 |
| minute | josch: thanks, do you see any other missing fields? | 20:58 |
| minute | hmm what's good about this is that i can maybe rerun just the build_qcacld2 + reprepro jobs after that | 20:58 |
| josch | minute: here are all fields of the *.deb of the qcacld dkms package which I uploaded to Debian: https://paste.debian.net/1313242/ | 21:01 |
| josch | Installed-Size is probably computed automatically | 21:02 |
| josch | minute: but you could download the *.deb from the last failed run and check | 21:02 |
| josch | yes, re-running the jobs after updating qcacld git might be enough | 21:02 |
| josch | success \o/ | 21:08 |
| josch | 6 and a half tries is a charm :) | 21:08 |
| minute | woohooo | 21:09 |
| minute | nice | 21:09 |
| minute | going for a walk with tina :3 | 21:09 |
| josch | tina has been waiting for CI finally being finished as well :D | 21:13 |
| + chomwitt (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 21:46 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@utdpat242060.utdallas.edu) | 21:59 | |
| - chomwitt (QUIT: Ping timeout: 268 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 22:14 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-fd68524e.019-141-67626730.bbcust.telenor.se) | 22:15 | |
| + bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 23:05 | |
| + minutepocket (~reform@ip5b40d69f.dynamic.kabel-deutschland.de) | 23:57 | |
| minutepocket | yo! | 23:57 |
| minutepocket | test driving pocket reform a bit. | 23:59 |
| josch | using the freshly built image? | 23:59 |
| minutepocket | i think it's from yesterday, and i rsynced the changes to emmc | 23:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!