| + anzu (~anzu@melkki.cs.helsinki.fi) | 00:01 | |
| - dodo (QUIT: Remote host closed the connection) (~dodo@user/dodo) | 00:36 | |
| + dodo (~dodo@user/dodo) | 00:36 | |
| - erle (QUIT: Ping timeout: 255 seconds) (~erle@2a02:8109:da40:c4:abea:939:4bb5:6b56) | 00:55 | |
| minute | strange https://source.mnt.re/reform/reform-debian-packages/-/jobs/1847#L4505 | 01:23 |
|---|---|---|
| - stites (QUIT: Ping timeout: 246 seconds) (~stites@130.44.147.204) | 01:46 | |
| - mtm (QUIT: Ping timeout: 246 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 02:02 | |
| + stites (~stites@130.44.147.204) | 02:12 | |
| - [tj] (QUIT: Server closed connection) (sid609767@id-609767.lymington.irccloud.com) | 03:04 | |
| + [tj] (sid609767@id-609767.lymington.irccloud.com) | 03:04 | |
| - nsc (QUIT: Ping timeout: 255 seconds) (~nicolas@192-99-142-46.pool.kielnet.net) | 03:25 | |
| + nsc (~nicolas@110-98-142-46.pool.kielnet.net) | 03:26 | |
| + erle (~erle@ip5f5bf355.dynamic.kabel-deutschland.de) | 03:30 | |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@64.189.201.150) | 03:49 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 04:10 | |
| + jacobk (~quassel@129.110.242.173) | 05:09 | |
| - klardotsh (QUIT: Ping timeout: 246 seconds) (~klardotsh@172.56.105.184) | 05:13 | |
| - jacobk (QUIT: Ping timeout: 246 seconds) (~quassel@129.110.242.173) | 05:16 | |
| - jnerula (QUIT: Ping timeout: 258 seconds) (~jnerula@li1009-93.members.linode.com) | 06:05 | |
| + jnerula (~jnerula@li1009-93.members.linode.com) | 06:07 | |
| josch | minute: libfiredecor.so was installed into debian/firedecor/usr/lib/*/wayfire | 07:15 |
| josch | but it's looking for it in debian/tmp/usr/lib/*/wayfire | 07:15 |
| - erle (QUIT: Ping timeout: 246 seconds) (~erle@ip5f5bf355.dynamic.kabel-deutschland.de) | 07:34 | |
| + erle (~erle@ip5f5bf355.dynamic.kabel-deutschland.de) | 08:06 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 08:25 | |
| - klardotsh (QUIT: Ping timeout: 255 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:06 | |
| + mjw (~mjw@gnu.wildebeest.org) | 09:12 | |
| - XYZ_ (QUIT: Remote host closed the connection) (~XYZ@78-80-106-89.customers.tmcz.cz) | 10:05 | |
| + XYZ_ (~XYZ@78-80-106-89.customers.tmcz.cz) | 10:12 | |
| + andreas-e (~Andreas@sauterelle.math.u-bordeaux1.fr) | 10:18 | |
| minute | josch: ah, strange | 10:19 |
| - mjw (QUIT: Quit: Leaving) (~mjw@gnu.wildebeest.org) | 11:33 | |
| + mjw (~mjw@gnu.wildebeest.org) | 11:34 | |
| - mjw (QUIT: Ping timeout: 246 seconds) (~mjw@gnu.wildebeest.org) | 11:38 | |
| * mark_ -> mjw | 11:47 | |
| + IchikaZou (~IchikaZou@45.8.223.248) | 12:08 | |
| - IchikaZou (QUIT: Remote host closed the connection) (~IchikaZou@45.8.223.248) | 12:09 | |
| - mtm (QUIT: Ping timeout: 246 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 13:18 | |
| - stites (QUIT: Remote host closed the connection) (~stites@130.44.147.204) | 13:22 | |
| + stites (~stites@130.44.147.204) | 13:22 | |
| - stites (QUIT: Remote host closed the connection) (~stites@130.44.147.204) | 13:22 | |
| + stites (~stites@130.44.147.204) | 13:23 | |
| minute | hmm, the firedecor dh_autoinstall receives an additional parameter, --destdir=debian/firedecor/, while this doesn't happen for wayfire | 13:24 |
| - robin (QUIT: Ping timeout: 248 seconds) (~robin@user/terpri) | 13:24 | |
| josch | minute: i guess you are somehow in a rush to have this implemented? | 13:31 |
| minute | josch: yeah, i want to ship it with RCM4 and it's also the base for the main pocket reform desktop | 13:33 |
| minute | josch: having a nice user friendly desktop that's still lean and doesn't require us to ship gnome and kde is a big win | 13:34 |
| minute | on X, there was xfce for this purpose, for example | 13:34 |
| minute | i think the only problem i need to understand/solve to get there is understanding where dh_auto_install arguments come from, i guess meson.pm | 13:35 |
| josch | meson doesn't drive dh, it's the other way round | 13:36 |
| josch | or does meson somehow have a "build .deb" backend? | 13:37 |
| minute | josch: yes, i know. i meant the meson module of debhelper | 13:37 |
| minute | josch: dh interpretes/drives the meson build | 13:37 |
| minute | interprets | 13:37 |
| josch | right | 13:37 |
| minute | and the "rules" for the two projects are identical, so my guess is its picked up from meson.build somehow | 13:38 |
| minute | ah yeah, lib/Debian/Debhelper/Buildsystem/meson.pm:167 | 13:38 |
| josch | minute: you say that you don't understand why d/rules of wayfire and firedecor are identical but the installation directory is different? | 13:42 |
| minute | josch: yes | 13:42 |
| josch | ah then i can explain | 13:43 |
| josch | wayfire builds multiple binary packages | 13:43 |
| minute | josch: oh? | 13:43 |
| josch | and firedecor only a single one | 13:43 |
| josch | the answer is in "man dh_auto_install" | 13:43 |
| josch | the default for --destdir is different depending on whether one or more than one binary packages are built | 13:43 |
| minute | josch: it also talks about compat levels... what is meant by that? i can see the perl code does it differently based on compat(13) | 13:44 |
| josch | you set the compat level via the build dependency on debhelper-compat (= 13), | 13:44 |
| josch | you already experienced how this is confusing so this is going to change with compat level 15 | 13:45 |
| minute | reading a bit in man debhelper-compat-upgrade-checklist | 13:45 |
| minute | eek | 13:45 |
| minute | > v13 This is the recommended mode of operation. | 13:46 |
| josch | yes | 13:46 |
| minute | ah, i see what you mean | 13:46 |
| minute | Build-Depends: | 13:46 |
| minute | debhelper-compat (= 13), | 13:46 |
| minute | cool | 13:47 |
| josch | in earlier times it was done via the file debian/compat | 13:47 |
| minute | i see | 13:47 |
| - stites (QUIT: Read error: Connection reset by peer) (~stites@130.44.147.204) | 13:48 | |
| minute | ok, another question. dh_install picks up paths from .install files. i thought these are the final paths as they should be on the target. but perhaps not? | 13:48 |
| minute | for example, i have in there > usr/lib/*/wayfire/libfiredecor.so | 13:48 |
| + stites (~stites@130.44.147.204) | 13:48 | |
| josch | yes, that sounds good | 13:49 |
| josch | what problem are you experiencing? | 13:49 |
| minute | dh_autoinstall puts them into debian/firedecor/usr/lib/x86_64-linux-gnu/wayfire though | 13:49 |
| minute | where dh_install cannot find them | 13:49 |
| josch | oh hm... that is strange... let me have a look... | 13:49 |
| minute | dh_install tries "debian/tmp" and ".." | 13:49 |
| josch | ah | 13:50 |
| josch | yeah | 13:50 |
| josch | sec... | 13:50 |
| minute | so there is a disagreement there, and i want to understand how i can either override the destdir path for dh_autoinstall or tell dh_install the right path | 13:50 |
| josch | try putting in the .install file the full path | 13:51 |
| josch | like: | 13:51 |
| josch | debian/firedecor/usr/lib/*/wayfire/libfiredecor.so | 13:51 |
| minute | ok! | 13:51 |
| minute | another naïve question: why is this building for x86_64? :0 | 13:51 |
| minute | or it's not really and that's just some artifact of the cross build? | 13:51 |
| josch | do you have a build log? | 13:52 |
| minute | https://source.mnt.re/reform/reform-debian-packages/-/jobs/1847 | 13:52 |
| josch | minute: your sbuild invocation is not making this a cross build | 13:54 |
| josch | you should add --host="$HOST_ARCH" | 13:54 |
| minute | josch: ah, i was copying from reform-tools | 13:55 |
| + mark_ (~mjw@gnu.wildebeest.org) | 13:55 | |
| josch | see build_patched.sh for examples | 13:55 |
| josch | minute: reform-tools builds an architecture-independent package, that's why it can be a native build | 13:55 |
| minute | ah :D | 13:58 |
| minute | cool btw, firedecor build worked now, locally | 13:59 |
| minute | now trying cross build. | 13:59 |
| josch | should work fine: http://crossqa.debian.net/src/wayfire | 14:00 |
| minute | ok, then i just need to fix something with my setup i think. on first try i get ERR_UNSOLVABLE because arm64 and amd64 compiler packages conflict | 14:01 |
| minute | cpp : Conflicts: cpp:arm64 but 4:13.2.0-1 is to be installed | 14:01 |
| minute | cpp:arm64 : Conflicts: cpp but 4:12.3.0-1 is to be installed | 14:01 |
| minute | etc | 14:01 |
| minute | or is that arm64/amd64 version mismatch? | 14:02 |
| josch | that sounds like you have different suites enabled? | 14:05 |
| josch | what can happen is a multi-arch version skew if a recent upload of a package like gcc happened and hasn't yet been built on all arches | 14:05 |
| minute | ok, i'm pushing to the server first to see what happens there | 14:06 |
| minute | josch: by different suites, do you mean for chdist vs the actual package or where could be a mismatch? | 14:08 |
| + robin (~robin@user/terpri) | 14:08 | |
| josch | minute: depends... do you run this locally on an amd64 box using the same scripts? | 14:08 |
| minute | josch: yes | 14:08 |
| minute | josch: i just ran custom_build.sh on my amd64 desktop | 14:09 |
| josch | then it should also fail on source.mnt.re -- let me have a look at the build log | 14:09 |
| josch | but first hanging laundry | 14:09 |
| minute | my desktop itself has "sid" | 14:09 |
| josch | since you are building with sbuild that should not matter | 14:09 |
| minute | alright. but the tar it uses for the filesystem matters perhaps? | 14:10 |
| josch | yes, that tarball can be set up in any arbitrary way of course | 14:10 |
| minute | ah yeah, it also breaks on the CI https://source.mnt.re/reform/reform-debian-packages/-/jobs/1852#L4058 | 14:11 |
| minute | so i have to build with the qemu method then i believe? | 14:12 |
| + murphnj (~murph@user/murphnj) | 14:18 | |
| minute | hmm, gcc for arm64 should also be 13.2.0 https://packages.debian.org/sid/arm64/gcc/download | 14:22 |
| minute | ah, i guess chdist is overwriting the apt sources | 14:26 |
| minute | ah, my mistake was to depend on g++ as a build dependency | 14:33 |
| - mark_ (QUIT: Ping timeout: 255 seconds) (~mjw@gnu.wildebeest.org) | 14:33 | |
| - ggoes (QUIT: Quit: ZNC - https://znc.in) (~gregf@fsf/staff/ggoes) | 14:34 | |
| minute | cross build goes brrr | 14:34 |
| + ggoes (~gregf@fsf/staff/ggoes) | 14:36 | |
| sevan | :) | 14:43 |
| josch | coming back and the problem has solved itself is the best :) | 14:52 |
| minute | hehe | 15:11 |
| minute | now fiddling with `dcmd` | 15:12 |
| josch | generally, you never need to build-depend on g++ because you get g++ automatically via build-essential | 15:12 |
| minute | josch: i see | 15:12 |
| minute | IIRC the .changes files are autogenerated next to the .debs yeah? | 15:14 |
| josch | yes | 15:14 |
| minute | mv: target '/home/minute/src/mref/reform-debian-packages/changes': No such file or directory | 15:15 |
| josch | but it exists? | 15:16 |
| minute | nope, ah i see this is normally created in .gitlab-ci.yml | 15:16 |
| minute | quite fiddly to get all of these details right, but almost there i guess | 15:17 |
| minute | i'm glad that debhelper includes automation for meson | 15:18 |
| + grimmothy_wares (~grimmware@2a03:b0c0:1:d0::efd:d001) | 15:19 | |
| - grimmware (QUIT: Quit: WeeChat 1.4) (~grimmware@2a03:b0c0:1:a1::6ad:1001) | 15:21 | |
| * grimmothy_wares -> grimmware | 15:21 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 15:25 | |
| minute | ~nice~ so the files are all there now in changes/ | 15:31 |
| minute | now to see if i can install them on my reform... | 15:31 |
| sevan | hmm, I wonder if Gnome 45 has landed in sid | 15:31 |
| sevan | ACTION updates reform | 15:32 |
| minute | sevan: i think not | 15:35 |
| minute | 44 still https://packages.debian.org/sid/gnome | 15:35 |
| minute | worked on the ci server too https://source.mnt.re/reform/reform-debian-packages/-/jobs/1864/artifacts/browse/repo/pool/main/w/wayfire/ | 15:36 |
| sevan | minute: ah, I have 44.3 installed currently | 15:37 |
| - grimmware (QUIT: Quit: WeeChat 3.5) (~grimmware@2a03:b0c0:1:d0::efd:d001) | 15:48 | |
| + marty_mcfly88 (~marty@2603-6000-8c01-0f3e-81b8-fee3-ee74-c36a.res6.spectrum.com) | 15:56 | |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 16:27 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 16:30 | |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@utdpat241106.utdallas.edu) | 16:32 | |
| minute | https://www.cnx-software.com/2023/09/05/raspberry-pi-cm4-compatible-risc-v-som-features-starfive-jh7110-soc/ | 16:38 |
| sevan | awesome! | 16:43 |
| sevan | 8GB model is only available without wifi, I don't know/understand the limitations with regard to PCIe, would it potentially be the same case that it would need a m.sata storage (instead of nvme). Would the existing ath9k card work? (on paper) | 16:51 |
| - mark_ (QUIT: Ping timeout: 246 seconds) (~mjw@gnu.wildebeest.org) | 16:59 | |
| q66 | i think the only limitation is that it's 1 lane of gen2, which is not much | 17:00 |
| q66 | i budged and got the visionfive2 last week, i want to try plugging a gpu in it for funsies | 17:00 |
| q66 | apparently it could work because it *should* be properly cache coherent | 17:01 |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 17:01 | |
| q66 | btw, the wifi module you get with vf2 is apparently just a usb dongle | 17:02 |
| q66 | so it's not worth talking about | 17:02 |
| sevan | so with 1 lane of gen2 you mean that you're not going to get full performance from a connected device, and there would be contention if there's more than 1 device on the bus? (if it's possible) | 17:03 |
| q66 | 1 lane of gen2 is 500MB/s | 17:03 |
| q66 | so that's your maximum throughput | 17:03 |
| sevan | ack | 17:03 |
| q66 | about the speed of sata3 bus | 17:04 |
| q66 | if you were to plug in an m.2 wifi module, it won't ever bottleneck you | 17:04 |
| q66 | it will bottleneck an nvme ssd, but it'll be usable | 17:04 |
| q66 | it'll likely bottleneck a gpu a lot | 17:04 |
| sevan | understood. | 17:04 |
| q66 | i'm still gonna try a gpu in it because it'll be funny | 17:05 |
| q66 | but first i gotta get my distro working on it properly | 17:06 |
| q66 | which means getting an appropriate kernel/u-boot combo packaged i guess | 17:06 |
| q66 | proprietary gpu userland driver is probably a lost cause because musl | 17:06 |
| q66 | but maybe at least display can work, with kernel modesetting? | 17:07 |
| q66 | and a separate gpu (if it works) would obviously work accelerated (but slower because of the pcie limitation) | 17:07 |
| sevan | genuine question, not being sarcastic: you're using musl by choice? | 17:07 |
| q66 | mostly by necessity, as there isn't really anything better | 17:08 |
| sevan | ack | 17:08 |
| q66 | glibc is a mess and does not work with clang+compiler-rt | 17:09 |
| q66 | it dlopens libgcc_s at runtime | 17:09 |
| sevan | glibc has a python build dependency (for running testsuite), I'll leave it at that. | 17:09 |
| q66 | musl is widely compatible in userland and works fine | 17:10 |
| q66 | it's not always ideal, but nothing is | 17:10 |
| sevan | I tried to build on Alpine long ago, they define error in header files they deemed you shouldn't use. annoyed me. | 17:11 |
| q66 | the main limitation has been poor allocator performance and i'm using a different allocator anyway | 17:11 |
| q66 | hm musl doesn't really do that | 17:11 |
| q66 | it doesn't have #error in any public headers | 17:12 |
| q66 | not sure if alpine does | 17:12 |
| q66 | but i think musl itself never did | 17:12 |
| sevan | what does your cdefs.h say? | 17:12 |
| q66 | musl does not ship cdefs.h and never did | 17:12 |
| q66 | on chimera that's provided by the musl-bsd-headers package for compatibility and definitely does not #error | 17:12 |
| q66 | maybe alpine's version has #error in it, i dunno | 17:14 |
| q66 | it has a #warning https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/libc-dev/sys-cdefs.h | 17:14 |
| q66 | i don't have even that because i find it pointless | 17:14 |
| sevan | that's nice. Asked Rich Felker about it at the time, was told I shouldn't be using cdefs.h so assumed it was via musl rather than alpine (gets tangled since Rich is/was associated with both?) | 17:15 |
| q66 | musl just doesn't come with the header, that's all | 17:15 |
| sevan | ack | 17:15 |
| q66 | it has to be supplied externally | 17:15 |
| q66 | i think dalias is an alpine user but does not contribute | 17:16 |
| q66 | but dunno, not very familiar with alpine's management | 17:16 |
| sevan | me neither. | 17:16 |
| sevan | left it there & moved on. | 17:17 |
| q66 | there are some people there i don't like (jirutka, mps), but for unrelated reasons | 17:17 |
| unixpoet | about the gen2 bottlenecking an NVME drive, I'd strongly suspect that the *CPU* on the VF2 would bottleneck before the bus would, it's not a very fast chip | 17:18 |
| minute | josch: wanna do a quick review? https://source.mnt.re/reform/reform-tools/-/merge_requests/45 | 17:18 |
| q66 | unixpoet: yeah, quite likely | 17:19 |
| q66 | i already have the system running on hifive unmatched, which has the same cores, and it's real slow | 17:19 |
| unixpoet | we already see that with eMMC IIRC, so NVME has no chance of seeing full speed even with a faster bus | 17:19 |
| q66 | around rpi3 level anyway | 17:19 |
| q66 | which is also slow :p | 17:19 |
| unixpoet | even rPi 4 is slow these days compared to what's out there | 17:19 |
| q66 | i'm building all the rv64 packages in transparent qemu-user emulation, as it's by far the fastest | 17:20 |
| unixpoet | yup. at one point there was a window where the fastest rv64 machine was qemu running on an M1 MacBook | 17:20 |
| q66 | and it's still annoyingly slow | 17:20 |
| q66 | i'm running it on a hetzner server with ryzen 5950x | 17:20 |
| unixpoet | or more precisely, an M1 Mini | 17:20 |
| q66 | you can kinda tell the ratios from the bars https://build.chimera-linux.org/#/waterfall | 17:21 |
| sevan | re M1 that's amazing. :) | 17:21 |
| unixpoet | I put Asahi on my M2 and zooooooooooom | 17:21 |
| q66 | x86_64 is i5-13600, riscv64 is 5950x + qemu-user -j32, ppc64le is power9 18-core baremetal, ppc64 is power9 16-thread virtualized, aarch64 is ampere emag 16-thread virtualized | 17:21 |
| unixpoet | the ampere is the VM host, or are you virtualizing an ampere guest? | 17:22 |
| q66 | vm host | 17:23 |
| unixpoet | I've been curious about their hardware because it seems to be the best-performing aarch64 hardware you can get outside of Apple right now | 17:23 |
| q66 | ppc64 and aarch64 run at osuosl | 17:23 |
| q66 | x86_64 and riscv64 run at hetzner | 17:23 |
| q66 | ppc64le is my own server in a local colo | 17:23 |
| minute | josch: my plan is to first merge https://source.mnt.re/reform/reform-tools/-/merge_requests/45 and then https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/14 (if pipeline succeeds), and then finally https://source.mnt.re/reform/reform-system-image/-/merge_requests/75 after doing some testing on a311d and imx8mq | 17:23 |
| unixpoet | is the power9 a Talos system? | 17:23 |
| q66 | yes | 17:23 |
| unixpoet | nice, you have lots of cool toys I want to play with :) | 17:24 |
| q66 | i also have a 2x18 core talos2 at home, but its bmc ethernet is fucked | 17:24 |
| unixpoet | aw :( | 17:24 |
| q66 | so it's not very usable as a server | 17:24 |
| q66 | but it's good as a local testing machine | 17:24 |
| unixpoet | indeed | 17:24 |
| q66 | my current workstation is ampere altra 80-core though | 17:24 |
| unixpoet | debian, fedora, something else? | 17:24 |
| q66 | chimera | 17:24 |
| q66 | i don't run third party distros | 17:25 |
| unixpoet | curious | 17:25 |
| q66 | well, the builders run a variety of distros, because there it does not matter much, they just need to run a buildbot worker and cbuild can run on any foreign systems since it containerizes its own build env | 17:26 |
| q66 | so these just run whatever they came with | 17:26 |
| q66 | aarch64 is ubuntu, ppc64 is debian, ppc64le is debian, riscv64 is ubuntu, x86_64 is ubuntu | 17:26 |
| q66 | but locally i don't run third party systems | 17:27 |
| sevan | q66: do you have a framework for maintaining builds of your distro? | 17:28 |
| q66 | as i said, buildbot | 17:28 |
| sevan | ah | 17:28 |
| q66 | it picks up git commits and builds everything for all archs | 17:28 |
| q66 | and uploads it to the master repo | 17:29 |
| sevan | so the logic for producing a new build is all done in buildbot. | 17:29 |
| q66 | e.g. lemme merge a pull request now | 17:29 |
| q66 | https://build.chimera-linux.org/#/waterfall and it goes | 17:30 |
| q66 | you can see the log running and all | 17:30 |
| josch | minute: i'll only be back online after ~22:00 because now is dinner time with baby but if that's not too late i'll review then | 17:35 |
| q66 | this time riscv64 finished faster than aarch64/ppc64 because unlike those two it's 1) built without LTO and 2) does not run tests (because running tests in a qemu-user env tends to be janky) | 17:35 |
| q66 | but otherwise it's much slower on average | 17:35 |
| minute | josch: that's fine, i'll try to test as much as possible manually first | 17:35 |
| sevan | q66: been nosing around, it's very impressive :) | 17:36 |
| q66 | obviously buildbot is just the instrumentation | 17:37 |
| q66 | i.e. it reacts to events from github, pulls the repo, then issues commands | 17:37 |
| q66 | all the actuall lifting is done by cbuild | 17:37 |
| q66 | cbuild is a result of 2.5 years of effort so it's quite comprehensive (the idea is to be self-contained and sandboxed so that it can run in any environment and produce identical output, and so that it can be used from any runner so that we don't get trapped in a specific environment with no way out) | 17:41 |
| q66 | it does containerized sandboxed building, it does repo management, signing, it gives packagers maintenance tools | 17:41 |
| q66 | it's also the fastest of all comparable systems that i know of | 17:43 |
| sevan | that's a lot of boxes to tick :) | 17:45 |
| q66 | also has no dependencies other than python (stdlib only) and bubblewrap | 17:45 |
| q66 | and apk | 17:45 |
| sevan | minute: riscv reform via cm adapter, pleeeeaaaaaase! :) | 17:47 |
| q66 | shouldn't it just work with the cm4 adapter | 17:47 |
| minute | theoretically yes, but needs device tree and driver hacks probably for DSI | 17:48 |
| q66 | funnily it should not be slower than the imx8mq | 17:48 |
| q66 | so it's potentially pretty viable | 17:48 |
| minute | q66: are you sure? i thought those cores are quite slow | 17:48 |
| q66 | they're about the same as cortex-a53, which is quite slow | 17:49 |
| q66 | the bad part is no open source gpu driver | 17:49 |
| q66 | as the thing has a powervr gpu | 17:50 |
| sevan | minute: that's good, so hardware wise we should be ok, but software wise we need to add support? | 17:50 |
| q66 | once i have a reasonable system with kernel modesetting working, i'm gonna check if i can perhaps coerce the proprietary driver into working by shimming any potential missing symbols by patchelf'ing the binary (and rewriting the NEEDED) | 17:51 |
| q66 | on nvidia driver this is not possible because of initial-exec TLS | 17:52 |
| q66 | whether the powervr driver uses initial-exec is to be seen | 17:52 |
| + grimmware (~grimmware@2a03:b0c0:1:d0::efd:d001) | 17:52 | |
| minute | sevan: yeah, DSI never works out of the box | 17:53 |
| q66 | i hope it does not require glvnd | 17:53 |
| minute | q66: ah, the open powervr gpu driver, is that for a different IP? | 17:53 |
| q66 | minute: the open driver is not ready, won't be for at least a year, and is vulkan-only | 17:53 |
| minute | q66: ah! vulkan-only may be ok with zink on top? | 17:54 |
| q66 | might | 17:54 |
| q66 | maybe in a year zink is in a shape where we can talk about that being realistic | 17:54 |
| q66 | last i tried running mutter/wayland with zink it would just hang :) | 17:54 |
| q66 | well, zink might as well be ready earlier than the powervr driver is | 17:55 |
| q66 | but yeah it's definitely not ready right now | 17:55 |
| minute | ok | 17:56 |
| q66 | i was experimenting with zink because the altra's pcie is messed up | 17:57 |
| q66 | https://github.com/chimera-linux/cports/blob/master/main/linux-lts/patches/9002-ampere-arm64-Work-around-Ampere-Altra-erratum-82288-.patch | 17:57 |
| q66 | https://github.com/chimera-linux/cports/blob/master/main/linux-lts/patches/9001-ampere-arm64-Add-a-fixup-handler-for-alignment-fault.patch | 17:57 |
| q66 | and i wanted to see if the vulkan driver had less display corruption | 17:57 |
| q66 | that was before i procured these patches | 17:58 |
| q66 | which i pulled from some tencent linux kernel fork | 17:58 |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 17:58 | |
| - andreas-e (QUIT: Quit: Leaving) (~Andreas@sauterelle.math.u-bordeaux1.fr) | 17:59 | |
| q66 | i don't think there is anything pending in the actual upstream to fix this, this is a major fuckup on ampere's part | 17:59 |
| q66 | but i'm fine potentially keeping the patches forever | 17:59 |
| q66 | i haven't noticed any perf degradation from the alignment fault handler | 17:59 |
| q66 | i guess because in practice it's almost never unaligned | 18:00 |
| q66 | arm and fucked up pcie implementations is like, unfortunately very common | 18:01 |
| unixpoet | can attest to that, the rockpro64 in particular is pretty picky | 18:11 |
| - stites (QUIT: Ping timeout: 246 seconds) (~stites@130.44.147.204) | 18:13 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@utdpat241106.utdallas.edu) | 18:14 | |
| sevan | hmm, 6.5 reform kernel landed and openzfs is not compatible (barfs at check during configure), 6.4 for now it is. | 18:14 |
| + stites (~stites@2607:fb91:dc3:f4f9:40b5:d12a:e4cd:feb2) | 18:19 | |
| + jacobk (~quassel@utdpat242094.utdallas.edu) | 18:22 | |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@utdpat242094.utdallas.edu) | 18:38 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 18:39 | |
| - mjw (QUIT: Killed (NickServ (GHOST command used by mark_!~mjw@gnu.wildebeest.org))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:53 | |
| * mark_ -> mjw | 18:53 | |
| + mark_ (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:54 | |
| - stites (QUIT: Read error: Connection reset by peer) (~stites@2607:fb91:dc3:f4f9:40b5:d12a:e4cd:feb2) | 19:09 | |
| + stites (~stites@130.44.147.204) | 19:09 | |
| - marty_mcfly88 (QUIT: Remote host closed the connection) (~marty@2603-6000-8c01-0f3e-81b8-fee3-ee74-c36a.res6.spectrum.com) | 19:30 | |
| + marty_mcfly88 (~marty@2603-6000-8c01-0f3e-a4bc-d69e-f2db-86ac.res6.spectrum.com) | 19:30 | |
| + chomwitt (~chomwitt@2a02:587:7a1b:5800:1ac0:4dff:fedb:a3f1) | 19:43 | |
| josch | minute: you don't need the x-prefix when doing string comparison in shell if you use quotes: https://www.shellcheck.net/wiki/SC2268 | 19:59 |
| minute | josch: ah oh :D i will remove it then | 19:59 |
| josch | interesting you think the change so great that it's now sysimage v4 :) | 20:03 |
| minute | josch: yeah, mainly because of the combination of the two images into one... | 20:04 |
| josch | you are right, that's a significant change | 20:04 |
| minute | and dropping a lot of packages (i still need to confirm that no essential packages are missing, i'm sure i missed something) | 20:04 |
| josch | <= back later for more comments | 20:05 |
| minute | cool, thanks josch | 20:09 |
| - pandora (QUIT: Ping timeout: 240 seconds) (uid585533@id-585533.ilkley.irccloud.com) | 20:13 | |
| + pandora (uid585533@id-585533.ilkley.irccloud.com) | 20:13 | |
| - chomwitt (QUIT: Remote host closed the connection) (~chomwitt@2a02:587:7a1b:5800:1ac0:4dff:fedb:a3f1) | 20:23 | |
| + IchikaZou (~IchikaZou@45.8.223.248) | 20:28 | |
| - IchikaZou (QUIT: Remote host closed the connection) (~IchikaZou@45.8.223.248) | 20:42 | |
| - Boostisbetter (QUIT: Ping timeout: 244 seconds) (4a410829d7@irc.cheogram.com) | 21:08 | |
| + jacobk (~quassel@129.110.242.224) | 21:13 | |
| - jacobk (QUIT: Ping timeout: 246 seconds) (~quassel@129.110.242.224) | 21:19 | |
| josch | minute: this new waybar config is what you now intend for both sway and wayfire? | 21:37 |
| josch | you wanted to test that it works with recent waybar | 21:37 |
| josch | i'm about to try it out myself | 21:37 |
| josch | (but not all of it because i have some other modifications) | 21:38 |
| minute | josch: yes, it works, it's not as nice as with an older version of waybar but it's fine | 21:38 |
| minute | josch: there was an old version of waybar where they used Buttons instead of Labels for the modules (except taskbar) and one could style their hover states to highlight them on mouseover. this is no longer possible | 21:39 |
| minute | it's not such a big deal right now though | 21:39 |
| minute | https://github.com/Alexays/Waybar/issues/1838 | 21:40 |
| josch | yes, i remember that and the long bug report on github where people complained about their broken configs :) | 21:40 |
| minute | yep, this was reverted unfortunately https://github.com/Alexays/Waybar/pull/1120 | 21:41 |
| minute | it was a good MR, one just would have to update one's config... but sigh | 21:41 |
| josch | reform-hw-setup contains a311d specfic stuff -- does this do anything strange on imx8mq? | 21:41 |
| minute | > If anyone is looking to retain buttons until Gtk4 migration is done or whatever, use commit ce8ae5bf17c11c441d8365acdb39625d32b6c2a2 | 21:41 |
| minute | josch: i think it should not do anything bad because those things will just error out on imx8mq | 21:42 |
| minute | because i.e. those alsa controls don't exist | 21:42 |
| josch | hm... maybe it would be cleaner to put an if/else or case/esac around those | 21:44 |
| josch | no idea whether some weird error messages might show up in the log, for example | 21:44 |
| minute | ok, can do | 21:45 |
| josch | i don't think the TARGET="$1" in reform-migrate is useful? | 21:45 |
| josch | what is the copyright/license for reform-mountains.jpg? | 21:45 |
| josch | my long-term plan is to upload reform-tools to debian, so it would be nice to have that kind of information documented :) | 21:46 |
| minute | josch: oh yeah. what would work for images in the context of debian, CC BY-SA 4.0? | 21:46 |
| josch | yes, that license is fine | 21:47 |
| minute | ok, i will put that in the README or...? | 21:47 |
| minute | wait, there's no readme | 21:48 |
| minute | josch: where would i put that? | 21:48 |
| josch | there is also no debian/copyright :D | 21:48 |
| josch | let me write something | 21:48 |
| minute | ok thanks! the image is (C) 2023 MNT Research GmbH / Philipp Broemme | 21:49 |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 21:50 | |
| josch | minute: the rest is gpl2+? | 21:50 |
| minute | v3+ i guess? | 21:51 |
| minute | josch: or are there compat. issues with v3? | 21:51 |
| josch | nope | 21:52 |
| josch | your choice :) | 21:52 |
| josch | reform-standby says gpl2+ | 21:52 |
| minute | josch: ah. can be bumped to v3+ | 21:52 |
| minute | josch: https://source.mnt.re/reform/reform-tools/-/blame/5ec005aef65353b7b082c82f5edef0e51890e46c/sbin/reform-migrate#L11 this says TARGET="$1" has been there from the beginning, so idk :D | 21:52 |
| minute | ah, you mean its twice in the script | 21:53 |
| josch | huh strange https://source.mnt.re/reform/reform-tools/-/merge_requests/45/diffs shows me a diff that says it got added | 21:54 |
| josch | minute: this debian/copyright is not complete yet but it's a start: https://paste.debian.net/1291058/ | 21:57 |
| minute | josch: super, thanks | 21:57 |
| minute | josch: i'll commit it, yes? | 21:57 |
| minute | i can also remove that TARGET line | 21:57 |
| josch | okay | 21:58 |
| josch | i didn't test your changes and just read the diff but i think it looks fine | 21:58 |
| josch | your changes to /etc/skel will not affect existing installations | 21:59 |
| josch | and the rest looks harmless | 21:59 |
| minute | josch: thanks a lot! suggested changes pushed | 22:00 |
| josch | cool :) | 22:00 |
| minute | incl copyright | 22:01 |
| minute | forgot to add the wallpaper to sway as well :D | 22:02 |
| josch | i'm now looking at https://source.mnt.re/reform/reform-system-image/-/merge_requests/75 -- that looks far more intricate | 22:04 |
| minute | josch: the a311d stuff you don't really need to review... just the changes to imx8mq would be worth to take a look | 22:05 |
| josch | didn't you want to add non-free-firmware? | 22:05 |
| minute | in any case we will use the generated imx8mq image for testing/installing a batch of new reforms starting tomorrow, so will probably find bugs and fix them | 22:05 |
| josch | for the realtek stuff i think? | 22:05 |
| minute | josch: yes, i added them only to a311d | 22:06 |
| minute | i'm not in a hurry to merge reform-system-image, but reform-tools and reform-debian-packages would be good to merge rather sooner than later, so i can use them in image builds | 22:06 |
| josch | reform-debian-packages passes so i guess that's that? :) | 22:08 |
| minute | alright :3 | 22:08 |
| minute | reform-tools merged | 22:10 |
| josch | i see the code in reform-system-image that builds upstream u-boot git with meson-g12b-bananapi-cm4-mnt-reform2.dts | 22:12 |
| josch | maybe ping vagrantc to ask how difficult it would be to let that be shipped by debian? | 22:13 |
| minute | josch: it needs binary TF-A BL2 though, need to check the license on that | 22:13 |
| josch | the proprietary bits come from https://github.com/LibreELEC/amlogic-boot-fip/tree/master/bananapi-cm4io right? | 22:14 |
| - stites (QUIT: Ping timeout: 246 seconds) (~stites@130.44.147.204) | 22:16 | |
| + stites (~stites@2607:fb90:ad60:d802:6a88:133c:2a05:da97) | 22:16 | |
| - stites (QUIT: Read error: Connection reset by peer) (~stites@2607:fb90:ad60:d802:6a88:133c:2a05:da97) | 22:16 | |
| + stites (~stites@c-71-232-14-73.hsd1.ma.comcast.net) | 22:17 | |
| minute | josch: yep | 22:19 |
| minute | same weird license as NXP stuff i guess https://github.com/LibreELEC/amlogic-boot-fip/blob/master/LICENSE | 22:19 |
| - stites (QUIT: Read error: Connection reset by peer) (~stites@c-71-232-14-73.hsd1.ma.comcast.net) | 22:19 | |
| josch | yes | 22:20 |
| josch | redistribution is prohibited | 22:20 |
| + stites (~stites@c-71-232-14-73.hsd1.ma.comcast.net) | 22:20 | |
| - marty_mcfly88 (QUIT: Read error: Connection reset by peer) (~marty@2603-6000-8c01-0f3e-a4bc-d69e-f2db-86ac.res6.spectrum.com) | 22:40 | |
| + marty__ (~marty@2603-6000-8c01-0f3e-a4bc-d69e-f2db-86ac.res6.spectrum.com) | 22:40 | |
| - jacobk (QUIT: Ping timeout: 246 seconds) (~quassel@utdpat241106.utdallas.edu) | 22:43 | |
| + jacobk (~quassel@129.110.242.224) | 22:44 | |
| - marty__ (QUIT: Read error: Connection reset by peer) (~marty@2603-6000-8c01-0f3e-a4bc-d69e-f2db-86ac.res6.spectrum.com) | 22:47 | |
| + marty_mcfly88 (~marty@2603-6000-8c01-0f3e-a4bc-d69e-f2db-86ac.res6.spectrum.com) | 22:47 | |
| - jacobk (QUIT: Ping timeout: 246 seconds) (~quassel@129.110.242.224) | 22:53 | |
| + IchikaZou (~IchikaZou@45.8.223.248) | 22:58 | |
| - IchikaZou (QUIT: Remote host closed the connection) (~IchikaZou@45.8.223.248) | 23:00 | |
| - mtm (QUIT: Ping timeout: 246 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 23:17 | |
| - kensanata (QUIT: Server closed connection) (~alex@user/kensanata) | 23:28 | |
| + kensanata (~alex@user/kensanata) | 23:28 | |
| - stites (QUIT: Ping timeout: 245 seconds) (~stites@c-71-232-14-73.hsd1.ma.comcast.net) | 23:30 | |
| + stites (~stites@2607:fb90:ad62:84d1:e06:3ee5:5b34:6465) | 23:31 | |
| - stites (QUIT: Read error: Connection reset by peer) (~stites@2607:fb90:ad62:84d1:e06:3ee5:5b34:6465) | 23:36 | |
| + stites (~stites@c-71-232-14-73.hsd1.ma.comcast.net) | 23:36 | |
| - Gooberpatrol66 (QUIT: Read error: Connection reset by peer) (~Gooberpat@user/gooberpatrol66) | 23:38 | |
| - mjw (QUIT: Ping timeout: 246 seconds) (~mjw@gnu.wildebeest.org) | 23:40 | |
| - iank (QUIT: Quit: ZNC 1.8.2+deb2+b1 - https://znc.in) (~iank@fsf/staff/iank) | 23:43 | |
| + iank (~iank@fsf/staff/iank) | 23:43 | |
| - iank (QUIT: Remote host closed the connection) (~iank@fsf/staff/iank) | 23:43 | |
| + iank (~iank@fsf/staff/iank) | 23:45 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!