gnu_srs1 | Howdy: Maybe supporting non-merged-user as default would be a win for Devuan: | 00:55 |
---|---|---|
gnu_srs1 | see https://lists.debian.org/debian-devel/2021/07/msg00126.html and replies, like: | 00:55 |
gnu_srs1 | the TC decision does indeed specify that we will implement merged-/usr using merged-usr-via-aliased-dirs. | 00:55 |
rrq | "aliased-dirs" ? | 00:56 |
golinux | Don't we already support non-merged-user? | 00:57 |
rrq | "aliased-dirs" -- is that kiddie speak for "links"? | 00:57 |
fsmithred | yes we support it and it is the default | 00:59 |
gnu_srs1 | So Devuan installations does not have merged-usr as default, good! But that is still an option? | 01:01 |
fsmithred | in expert install you get a checkbox | 01:04 |
fsmithred | "no" is pre-checked | 01:04 |
gnu_srs1 | Maybe we should make a difference her as compared to Debian, that would make more users to adopt Devuan? | 01:08 |
gnu_srs1 | see e.g. https://lists.debian.org/debian-devel/2021/07/msg00128.html | 01:08 |
gnu_srs1 | See also: https://lists.debian.org/debian-devel/2019/07/msg00498.html | 01:15 |
rrq | is this now outdated? https://www.debian.org/releases/buster/amd64/apcs02.en.html | 01:36 |
rrq | and this is not applicable to debian anymore? https://refspecs.linuxfoundation.org/FHS_3.0/fhs-3.0.html | 01:39 |
rrq | also outdated? https://wiki.debian.org/FilesystemHierarchyStandard | 01:41 |
mason | Random note, I just noticed that #debian-derivatives exists on OFTC, and I've added it to my list of haunts. I don't know if it'll be useful to us or not. | 04:39 |
bgstack15 | Well, Debian abandoned LSB, so why not FHS as well? | 05:21 |
bgstack15 | And actually based on that page rrq linked https://www.debian.org/releases/buster/amd64/apcs02.en.html, "It is now recommended to have /usr on the root partition /, otherwise it could cause some trouble at boot time." | 05:21 |
bgstack15 | So in other words even /usr (presumably for /usr/bin and /usr/lib) is essential, so there's no difference between /bin and /usr/bin, to Debian. | 05:22 |
bgstack15 | But then I'm one of those hacks who hasn't cared too much about UsrMerge one way or another. I wish for stability, but I guess I know where to look if for some reason my $PATH is busted. | 05:23 |
bgstack15 | But in the discussion today we debated the policy of Devuan. I didn't hear, "We are Debian but without UsrMerge." | 05:24 |
yeti | is that no longer optional in the installer? | 05:27 |
yeti | deBIan's installer? | 05:27 |
rrq | no, not in debian installer | 05:31 |
mason | yeti: IT's available in debootstrap, but not the installer itself. | 06:15 |
yeti | not even in expert mode? | 06:49 |
yeti | then "We are Debian but without UsrMerge." should be put very near to the without-systemd-line | 06:49 |
yeti | what di they smoke? diesel exhausts? | 06:50 |
yeti | o | 06:50 |
golinux | It was at the top of the Release Notes when it was first implemented. | 06:52 |
yeti | I always thought it were at leat optional in expert mode of the deBIan installer... but maybe I only dreamed it... | 06:57 |
rrq | note for buster netinstall; you'll have to go C-A-F2 and create /target/{bin,sbin,lib,lib32,lib64,libx32} by hande before installing the base system | 07:13 |
rrq | (too many e e) | 07:13 |
mason | https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=978636 | 07:15 |
yeti | nuts! if at all the FS gets redefined, it should even be less eggs in one bag... | 07:16 |
mason | I feel like I'm missing something. Here: https://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge/ | 07:19 |
mason | They say: "With the merged /usr directory we can offer a read-only export of the vendor supplied OS to the network, which will contain almost the entire operating system resources." | 07:20 |
mason | I'm guessing the intent would be for the initramfs to contain all the moving parts for mounting this up, and if there's any sort of network disruption, game over. | 07:20 |
mason | ...as opposed to the reason why there's a split in the first place, which allows for a cohesive minimal system that can access network resources on its own. | 07:21 |
mason | The other funny bit is "Myth #6: A split /usr is Unix “standard”, and a merged /usr would be Linux-specific" | 07:22 |
mason | Their answer is this: "Fact: On SysV Unix /bin traditionally has been a symlink to /usr/bin. A non-symlinked version of that directory is specific to non-SysV Unix and Linux." | 07:22 |
mason | The funny thing there is that *they're also moving /sbin* which invalidates that reasoning entirely. | 07:22 |
mason | </rant> | 07:23 |
parazyd | fsmithred: So I would say it's because of this: https://git.devuan.org/devuan-sdk/live-sdk/src/branch/master/config#L19 | 09:40 |
parazyd | If you didn't change it to i386 | 09:40 |
fsmithred | parazyd, thanks. It looks like it's working now. (squashing) | 13:24 |
fsmithred | Will it work if I move that into the blend config, where I already have to change stuff for each arch? Or does it have to be sourced before the blend? | 13:25 |
parazyd | fsmithred: I'd recommend just changing the config file $arch locally depending on what you're building | 13:28 |
fsmithred | which config file? | 13:28 |
parazyd | I could rework it a bit, but zero time atm | 13:28 |
parazyd | live-sdk/config | 13:28 |
fsmithred | I might try moving it into blend config | 13:29 |
fsmithred | will I have to move the test for kernel choice, too? | 13:29 |
fsmithred | it's at the bottom | 13:29 |
fsmithred | actually, I think the answer to that is yes. I did move it into blend config but not out of live-sdk/config. | 13:30 |
fsmithred | so the kernel was chosen before the arch definition was changed. | 13:30 |
parazyd | Yeah | 13:37 |
parazyd | Exactly | 13:37 |
fsmithred | Nope: [W] load(): required setting is blank: arch | 13:58 |
parazyd | This is in line 51 in live-sdk/sdk | 14:00 |
parazyd | One sec | 14:00 |
parazyd | fsmithred: http://sprunge.us/iaLXp6 | 14:01 |
parazyd | This patch should work for your usecase | 14:01 |
parazyd | Along with the config file changes | 14:01 |
fsmithred | thanks | 14:02 |
fsmithred | does that need to be in the .blend file or can it be in the blend config file? | 14:06 |
parazyd | This is in the sdk file (the one that you source) | 14:08 |
parazyd | cd path-to-live-sdk && curl http://sprunge.us/iaLXp6 | patch -p1 | 14:08 |
fsmithred | yeah, and the req gets moved to another file | 14:08 |
parazyd | No | 14:08 |
fsmithred | ? | 14:08 |
parazyd | It'll first just check for $os | 14:09 |
fsmithred | so just remove 'arch' | 14:09 |
parazyd | ah nvm I made a mistake | 14:09 |
parazyd | Yeah you can just remove the $arch check | 14:09 |
fsmithred | (*) Done! Thanks for being patient! | 15:23 |
parazyd | :) | 15:26 |
fsmithred | I modified the test so that I have to comment/uncomment a line so that minimal-live gets 686 and desktop-live gets 686-pae. It's better to have that in my config file so I don't have to change the top-level one. (But I did anyway 'cause some package is missing in chimaera.) | 15:32 |
fsmithred | ok, and that package is back in chimaera. firmware-linux | 15:32 |
fsmithred | looks like we should not install firmware-linux by default | 15:35 |
fsmithred | #arch="amd64" | 15:35 |
fsmithred | arch="i386" | 15:35 |
fsmithred | case "$arch" in | 15:35 |
fsmithred | # for minimal-live | 15:35 |
fsmithred | #i386) extra_packages+=(linux-image-686) ;; | 15:35 |
fsmithred | # for desktop-live | 15:35 |
fsmithred | i386) extra_packages+=(linux-image-686-pae) ;; | 15:35 |
fsmithred | amd64) extra_packages+=(linux-image-amd64) ;; | 15:35 |
fsmithred | esac | 15:35 |
fsmithred | oops | 15:35 |
fsmithred | wrong paste, sorry | 15:35 |
fsmithred | firmware-linux Depends: firmware-linux-free, firmware-linux-nonfree (= 20190114-2) | 15:35 |
fsmithred | chimaera desktop-live test isos are up: https://files.devuan.org/devuan_chimaera/desktop-live/ | 17:29 |
fsmithred | or use your favorite iso mirror | 17:29 |
gnu_srs2 | Question: Do the ISOs have non-merged-/usr as default and an option to use merged-/usr when entering the expert menu? | 17:44 |
fsmithred | the live isos do not have merged /usr and do not give you the option | 17:45 |
fsmithred | the installer just copies the live system to hard drive and makes it bootable | 17:46 |
fsmithred | it's refractainstaller, so there's no expert install (or there's no non-expert install, depending on your POV) | 17:46 |
gnu_srs2 | fsmithred: That's OK with me. What about the other ISOs? | 17:47 |
fsmithred | the installer isos give you a choice in expert and choose non-merged for the non-expert. | 17:47 |
gnu_srs2 | Good :) | 17:48 |
fsmithred | I should say 'default to non-merged' in regular install. | 17:48 |
gnu_srs2 | Debian is loosing again... | 17:48 |
fsmithred | Too bad, but we always need to blood, and that'll help. | 17:48 |
fsmithred | to blood/NEW blood | 17:49 |
golinux | (fsmithred is picking up my typing habits) | 18:00 |
fsmithred | my fingers are sneaking past my eyes and doing their own thing. | 18:01 |
golinux | Welcome to my world. Sometimes I type word salad . . . | 18:01 |
golinux | It doesn't get better. | 18:02 |
fsmithred | the only advantage to not seeing is that when you type word salad, I can usually read it. | 18:09 |
fsmithred | golinux, checkout the desktop-live iso. I was able to patch desktop-base in the build, so the menu icons are back. | 18:17 |
fsmithred | afk, bbl | 18:25 |
golinux | Will do that shortly! | 18:53 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!