| freem | it's mixed apparently | 00:00 |
|---|---|---|
| freem | which is another weird shit debian does with them | 00:00 |
| freem | https://p.mort.coffee/Fof.png | 00:01 |
| freem | so the normal repo does not contains dbgsyms for those, at least | 00:01 |
| freem | I had them installed prior to my move to devuan | 00:01 |
| rrq | tell me the package you need | 00:01 |
| freem | they were located in that stuff I linked above | 00:01 |
| rrq | one is enough | 00:02 |
| rrq | I'm not following random links | 00:02 |
| freem | libgcc-s1-dbgsym libgmp10-dbgsym libhogweed6-dbgsym libnettle8-dbgsym libsdl2-2.0-0-dbgsym libstdc++6-dbgsym zlib1g-dbgsym | 00:02 |
| freem | random links to debian's official documentation showing that it does have specific repo for those? | 00:02 |
| freem | I see. | 00:03 |
| gnarface | freem: devuan hasn't quite replicated 100% of debian's server infrastructure. i haven't heard of there being a special repo for debug packages, but we don't have "snapshots" or "experimental" either | 00:03 |
| freem | I'm using stable | 00:03 |
| gnarface | maybe ask fsmithred | 00:03 |
| freem | too old for the funny stuff | 00:03 |
| gnarface | but the answer might just be that we don't have the time or space | 00:03 |
| freem | I grew lazier as I grew older :) | 00:03 |
| freem | well, that'd be sad, because it's a deal breaker for me. debug symbols are important tools. | 00:03 |
| freem | for my use case, that is. | 00:04 |
| gnarface | i'm pretty sure those packages are easy to make... | 00:04 |
| freem | oh, they are | 00:04 |
| freem | but then, if I rebuild packages, it would be simpler and faster for me to just recompile devuan's KDE tools I need, which made me migrate, and isntall them on debian | 00:04 |
| gnarface | it might just be an oversight too, maybe it would work but just isn't in amprolla because nobody needed them yet | 00:04 |
| freem | which is fully documented on that topic, and will reduce friction when I'll mention the distro I use in a push request or a bug report | 00:05 |
| freem | basically, the only tools I have which want hard to have systemd installed are those 3 KDE tools: kcachegrind, massif-visualizer, and hotspot | 00:06 |
| freem | they are even "only" convenience tools, I technically could do the work without them... but I'm not masochist | 00:07 |
| freem | especially for massif, the text output is non-trivial to read, with all the STL template mess | 00:07 |
| buZz | i wish https://github.com/linux-surface/linux-surface didnt depend on systemd | 00:08 |
| buZz | or that i was more fluent in removing that ;) | 00:08 |
| freem | it's just a distro to run on a MS surface device? | 00:10 |
| buZz | its a modified kernel with some libraries and tools replaced with newer and/or patched versions | 00:11 |
| freem | it says it have a special kernel (probably for drivers?), but I doubt their kernel build depends on systemd? | 00:11 |
| fsmithred | freem, does debian have the dbgsym packages you want? | 00:11 |
| freem | yes it does | 00:11 |
| freem | otherwise how could I have them installed? | 00:11 |
| fsmithred | are they in a repo separate from the main ones? | 00:11 |
| freem | yes | 00:11 |
| rrq | not in their repo | 00:11 |
| buZz | freem: its mostly the intel wacom touchscreen stuff, for multitouch and pen interaction | 00:11 |
| freem | as said above, with the links to their documentation expliciting this | 00:11 |
| buZz | (i think) | 00:12 |
| fsmithred | I just looked at libhogweed6 and we don't fork that package. | 00:12 |
| freem | buZz: then you could just build that part and install it on devuan? | 00:12 |
| fsmithred | The one that's in our repo isn't actually "in" our repo. | 00:12 |
| buZz | perhaps ;) | 00:12 |
| rrq | freem: you need to download those from the debian points then | 00:12 |
| freem | I understand that you don't fork all of them, fsmithred, but I refuse to take the risk of fucked up dbgsyms because of a mismatching lib/sym | 00:12 |
| fsmithred | exactly | 00:12 |
| freem | so, is there is a list of recompiled libs? | 00:13 |
| fsmithred | uh, get the same version of dbgsym as the package itself | 00:13 |
| freem | is glibc6 patched, for exampled? | 00:13 |
| freem | example* | 00:13 |
| fsmithred | you mean do we fork it? | 00:13 |
| freem | yes | 00:13 |
| fsmithred | no f'n way | 00:13 |
| freem | but I still need a full list, to ensure no lib I have in the binaries is a modified one | 00:13 |
| freem | and this, through the whole dependency graph | 00:14 |
| fsmithred | dpkg -l |grep devuan | 00:14 |
| fsmithred | will show installed packages that we forked | 00:14 |
| freem | it's the only way I could trust those, and no way I'll toy with risky stuff there, debugging is already painful enough | 00:14 |
| rrq | the version code for forked packages includes "devuan" | 00:14 |
| freem | see | 00:14 |
| fsmithred | do you understand the basics of how amprolla works? | 00:14 |
| freem | libdbus | 00:14 |
| freem | libsdl depends on libdbus | 00:15 |
| freem | which means no program loading sdl can use debian's symbols for devuan | 00:15 |
| freem | what's amprolla? | 00:15 |
| fsmithred | the merging software that sucks packages from debian and makes it look like they come from devuan servers | 00:16 |
| freem | thanks. You got your answer, too :) | 00:16 |
| fsmithred | and filters out systemd and anything that depends on it | 00:16 |
| freem | well then, if there's no clean, official way to get dbgsyms from devuan installed, I guess I'd better go back to debian and only compile the 3 tools I need, seems the path of least resistance | 00:19 |
| rrq | cheers | 00:19 |
| freem | yay. | 00:19 |
| freem | or I can just frankenbian as I did in good old times | 00:20 |
| rrq | you don't need to tell here | 00:20 |
| freem | yup, seems even easier, and since I don't debug kde stuff anyway, globally safe | 00:20 |
| freem | indeed | 00:20 |
| freem | I won't mention debugging tools here anymore :) | 00:20 |
| fsmithred | I do see dbgsym packages for dbus, but I don't know if that's enough. (above my pay grade) | 00:23 |
| freem | don't worry, I'll just tinker my way | 00:23 |
| freem | reading apt pinning's doc anew | 00:23 |
| freem | can't say I was missing that tool, but it *is* handy in some situations | 00:24 |
| fsmithred | lol, yeah. been there, done that. | 00:24 |
| Xenguy | freem, https://www.devuan.org/os/packages#debug-symbols | 00:49 |
| freem | I see | 00:50 |
| freem | well, I'm way too wary of the problems this kind of mix can generate. Debugging is already painful enough, I don't want to even take the risk of countless hours debugging the wrong symbols | 00:50 |
| freem | but thanks for the actually useful link | 00:51 |
| freem | instead of repeating that everything is in the same repo, despite pointers at evidence that nope. | 00:51 |
| Xenguy | You're most welcome. I wish I could help more, but it's above my paygrade. The debug section I linked was added by a Devuan wizard, but that's about all I can tell you on the topic | 00:52 |
| freem | no problem | 00:52 |
| freem | there is no problem in not knowing. What I don't like though, is when someone who does not knows refuses to follow links to official doc of the forked project, that the fork mirrors, and says the contrary of what doc says | 00:53 |
| freem | that is... a bit frustrating. | 00:53 |
| Xenguy | Best of luck | 00:54 |
| freem | thx | 00:54 |
| freem | it won't be my 1st frankenbian | 00:54 |
| freem | oh, btw, the eudev package can't be deinstalled, I have not digged the topic yet, but it is likely the post/pre-rm/inst scripts in it are buggy | 00:56 |
| freem | I'll dig in the topic when I'm finished moving back. Is this place interested at the findings I might have? | 00:56 |
| freem | as a reminder, eudev is not technically necessary to run a distro, so it should be possible to remove it | 00:57 |
| gnarface | doesn't the udev package have the same problem? i thought if you remove either of them without replacing it with the other one it removes the kernel too, or something like that... | 00:57 |
| freem | IIRC it does not | 00:58 |
| freem | but I can try | 00:58 |
| freem | actually, I *shuold* try, since there's no point in running (e)udev on my VPS | 00:58 |
| freem | my toy VPS, that is. A perfect target to tinker with udev-less system | 00:58 |
| freem | but if udev had the same problem, then it would still keep some config files after migrating toward devuan | 00:59 |
| freem | I noticed that when I did a back&forth, few days ago, and forgot to mention it | 01:00 |
| freem | it's nothing I can't handle, trenchbroom's package had similar kind of bug, because it was expecting a package which exists and is mandatory on ubuntu but I had not installed it | 01:00 |
| freem | I know the stuff enough to debug and clean mistakes in packages though, so I reported the problem, fixed it, and they implemented something derived from my patch. | 01:01 |
| gnarface | we do have a bugs.devuan.org, for devuan-forked packages, eudev might count... | 01:02 |
| gnarface | whether they'll fix it or not i don't know, but if it's doing something the main udev package doesn't, that increases the chances | 01:03 |
| freem | I'll investigate quickly, since I'll need to clean my system anyway | 01:03 |
| AlexLikeRock | hiiiii ! | 01:07 |
| golinux | freem: FYI . . . https://git.devuan.org/devuan/amprolla3 | 01:11 |
| golinux | Hi AlexLikeRock !! | 01:11 |
| freem | thanks, could be handy someday | 01:12 |
| AlexLikeRock | hi golinux :-D | 01:13 |
| freem | so, about de-installing udev (the systemd one) on debian, it *is* possible. The kernel (linux-image-6.1.0-25-amd64 and other variants) depends on either initramfs-tools or linux-initramfs-tool, which are provided by either initramfs-tools, dracut, or tiny-initramfs. tiny-initramfs depends solely on cpio, which means no udev enforced by dpkg (and thus, apt, apt-get, dselect, aptitude, name your favorite frontend). This, in turn, means that devuan | 02:27 |
| freem | requires more tinkering around if you want to get rid of udev entirely, which can be a very valid thing on various situations including embedded systems or VMs (when you want them *really* lightweight, especially at boot time). | 02:27 |
| freem | it's a bit unfortunate that there's no way to get kernels booting directly /, but I guess this kind of stuff is both almost lost knowledge, and not supported for any systemd-based system, since might require to not do the usrmerge thing | 02:28 |
| freem | in any case, this leads to the fact it is impossible to purge eudev from the system a regression compared to debian | 02:29 |
| freem | ofc, this is a specific use case, but on that one, debian is indeed closer to the goal of being universal | 02:29 |
| freem | as for the bug which prevents the purging, apparently eudev expects /var/log/udev to be a file or a symlink. In my case, because I am running my own init with own scripts, it is a folder, which made eudev post-rm scripts unhappy | 02:30 |
| freem | took me some time to repair my system. I guess I should rewrite my /etc/sv/udev/run script in a way it can handle both variants... or just get rid of the udev need, as I almost never plug external stuff anyway | 02:31 |
| freem | (but I was always too lazy to push my knowledge far enough to be able to do that) | 02:32 |
| freem | that is all my findings on that topic (and I consider the case close, since it seems I'm on myself for stuff as "specific" as installing debugging tools that I can trust won't make waste the time I dedicate to contributing to open source projects.) | 02:34 |
| * freem now go building local packages for the 3 kde tools he needs | 02:35 | |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
