libera/#devuan-dev/ Monday, 2019-10-07

fsmithredrrq, where do the menuentry lines for grub.cfg come from?05:11
fsmithredI can find all the .cfg files for isolinux, but boot/grub.cfg doesn't have boot entries05:11
rrqnot sure :)05:11
fsmithredgrep says so05:13
fsmithredin one of the earlier isos, I booted uefi mode and it didn't have all the boot choices05:13
fsmithredwill try with the newer isos05:13
fsmithredI did a no-network install from dvd-1 today05:15
fsmithredit worked05:15
fsmithredupgrade to beowulf was easy and flawless05:16
rrqyes I haven't done things to the uefi boot option; maybe it needs fixing05:17
fsmithredlooks like they're all there05:20
fsmithredplus a few i've never seen before05:20
fsmithredthere's something called Desktop environment menu05:21
rrqI think it's got "polluted" with some debian hacks...05:21
fsmithredand in it: Gnome desktop boot menu, KDE desktop boot menu, LXDE...05:21
fsmithredjust those three05:22
rrqthe extlinux boot is clean and simple05:22
rrqmaybe there's a configuration somewhere.. otherwise I'll need to do perl patching05:23
fsmithredeach of those items just presents the usual Install options, but says it's for that particular desktop05:24
rrqmmm I think it shouldn't be there at all05:25
fsmithredoh, no Graphical install options05:25
fsmithrednope. graphical install is not hidden in the desktop install options05:26
rrqI'll need a qemu variant with uefi ...05:29
fsmithredoh, I couldn't get a boot menu with qemu05:36
fsmithredjust the tiano core stuff05:36
fsmithredI'll try again.05:36
fsmithredon amd64 the screen is tiled and unreadable05:39
fsmithredI can see that it's on the boot menu, so I pressed e and added 'set gfxpayload=1024x768' and it's good05:42
rrqok. should have that added05:44
fsmithredit works ok in virtualbox05:44
fsmithredI've seen this before, with my live isos05:48
rrqi'm now trying the "ovmf" boot with qemu05:49
fsmithredI think I hard-coded 640x480 somewhere05:49
fsmithredyeah, that's what I use05:49
fsmithredon amd64 dvd105:49
rrqlooks kind of fine except that it affects the start-up terminal first ?? then brings up the vm window but without the graphical install options05:50
rrqthe start-up terminal goes black05:50
rrqwith cursor up in home corner05:50
fsmithreddid you get to Shell> prompt?05:51
rrqno prompt, but a grub boot options screen  with three options05:52
rrqnot tiled or anything05:52
fsmithredusing any graphics options in your qemu command?05:52
rrqwell all same as before, but I added the -bios option...05:53
fsmithredI tried -vga std and -vga qxl05:53
rrq-vga qxl -serial mon:stdio -echr 0x1c05:53
rrqI wonder if one could run extlinux/isolinux through grub as well>05:55
fsmithredwhat do you mean "through"?05:55
fsmithredbtw, that worked. I got the boot menu.05:56
rrqmm it's "-serial mon:stdio" that makes the start-up terminal involved in grub05:57
rrqwith ^\c I can shift it to monitor mode05:57
fsmithredI'll take your word for it05:57
onefangMy Magic Pixie Dust thing chains to extlinux from grub for some of the OSes on it.05:58
onefangEr from grub 2.05:58
rrqfsmithred: "trough" meant "instead of" actually :)06:36
rrqlet me first explore the variant of adding the gtk option(s) .. I think I found the gnit to squeeze06:37
fsmithrednot sure which of the syslinux tools would work best (or at all) for uefi boot06:37
fsmithredone thing I'm also concerned about is that the isolinux files are all in the root instead of /isolinux06:38
fsmithredat the very least, it's messy06:38
fsmithrednot sure if it breaks anything06:38
fsmithredthe mini.iso is that way06:38
rrqyeah this building is a mini iso with the injection of a pool06:39
rrqI was thinking noone sane would look at the installer file structure :)06:40
fsmithredwell, that's where they will find the pretty docs06:40
rrqtrue ... but who reads docs nowadays anyhow ..06:41
fsmithredwe have pictures06:41
rrq:) ok maybe I should look at making it neater06:41
fsmithredonce we get this done, it's going to be easy to do the same with beowulf, right?06:42
rrqyes that's the idea.06:43
rrqI started on those build branches already06:43
fsmithredI should probably try a run with live-sdk in beowulf. The desktop would probably work now.06:44
rrqI should probably drop in a HOWTO.adoc as well just for future forensics06:45
golinuxI am finishing up the iso-docs but need the exact address of the torrent for the point release.   Would "devuan_ascii_2.1.torrent" work?10:22
golinuxjaromil: parazyd: ^^^10:23
parazydgolinux: Yes10:53
parazydThough if you want to stick to our versioning that we had so far, it should be
masonSo, is there a task list noting things in the way of a Beowulf release?15:02
golinuxparazyd: Thank you for responding.  That is the versioning that fsmithred et al decided on.  I questioned it as being inconsistent but . . . Unfortunately, it is probably in too many places to think of changing at this point,17:31
golinuxmason: No.  Just what has been discussed at meets here and on the -dev list.17:33
LeePenI have just done a fork of udisks2 from buster for beowulf without libpam-* dependency.17:35
LeePenWith some updates to policykit dependencies it should be possible to use either elogind or consolekit  with udisks2.17:35
masongolinux: Might be worth having an official list on the web site, with the notion that it might inspire people to pick off tasks and push us closer.17:36
LeePenProposed trees for review posted to devuan-dev17:36
golinuxmason: Feel free to make one if you think it would be helpful.17:37
masongolinux: Not useful. I don't know what's left thoroughly enough.17:38
masongolinux: I'd be in a good position to drill down on specific tasks, though, and a list would draw out others who'd feel similarly.17:38
LeePenmason: you could review and test my udisks2 and policykit?17:39
masonLeePen: Are they for ASCII or Beowulf? I'd need to spin up a Beowulf if the latter.17:39
golinuxYou could go through the list of outstanding bugs at and post the ones critical to the release.  That has always been the most useful list17:40
masongolinux: I was looking at that earlier and it seemed to lack a focussed set related to Beowulf. I'll find an example of what I'm thinking about.17:40
masongolinux: Here's an example:    The bug tracking system there talks about different branches, and tracks bugs tagged for particular branches.17:43
masongolinux: Even more relevant:
masongolinux: Do we tag bugs as blockers?17:45
masonLeePen: Are there patches beyond Message-ID: <> or is that the full set?17:47
golinuxThose deemed critical are blockers.  If you search this list you will find that KatolaZ did that for the ASCII release.  You'd have to sort through every one and make a decision.17:50
LeePenmason: Message-ID: <>17:51
masonOh, even easier.17:51
LeePenThey will need to be taken together.17:52
LeePenpolicykit-1 can go through unstable, but udisks will probably have to go straight into beowulf as Debian's unstable has moved on.17:52
masonLeePen: And do you have a notion of what you want tested? I assume things install, so is there behaviour to test? udisks2, the only thing that pops to mind is automounting of hotplug media.17:52
LeePenYes. I don't have a bare metal install of beowulf, so I can't actually test media insertion etc.17:53
LeePenIf you can test with both consolekit and libpam-elogind doing session/pid matching, that would be great. Thanks.17:54
masonThat can be done via VM as well, but yeah. I can see about a bare metal install and try some of this. I had to do a bunch of udev device insertion debugging a few months ago, so I still have my extension cable so I don't wear out my USB port. :P No udisks2 there. I dislike udisks2 fairly intensely. But I'll install it for this.17:55
masonLeePen: What desktop environment do I want to install for it? I don't use one normally.17:55
LeePenBoth trees compile and the resulting packages install correctly for me.17:55
LeePenI don't use it either, but people have been talking about sorting this, so I thought I would do it.17:56
masonSo, 1) install Beowulf, 2) build packages from those, 3) install, 4) exercise udisks2.17:56
masonI imagine xfce4 would suffice.17:56
LeePenI think the target is that it will be DE agnostic (and elogind|consolekit) agnostic.17:56
masonOh, you know, I'll take whatever the "Desktop" task gives me.17:56
masonYeah. Just have to pick one to start.17:57
LeePenYes xfce4. gnu_srs1 was using MATE IIRC.17:57
LeePenIt might even work without one.17:57
LeePenThanks. bbiab18:00
amesserjust used reportbug ending up with a bug report for apt in devuan bug tracker20:20
amesserbut apt is maintained at debian...20:20
LeePenreportbug -B debian is your new friend! ;)20:22
amesserlo, yes20:22
masonLeePen: Do you know offhand if we have install media that defaults to Beowulf, or am I installing ASCII and upgrading to test?20:24
amesserafaik there are no install media for beowulf available at the moment20:26
masonUpgrade it is, then.20:26
amesserwould be simplest and safest20:27
amesserjust install minimal ascii20:27
amessersometimes it is possible to use debootstrap but this required additional manual work afterwards (which is otherwise done by the installer)installer)20:28
masonReasonable start.20:29
masonOh, sorry, was in scrollback.20:29
masonEh, I'm happy upgrading. Minimal install, as you note, will go fairly fast.20:29
masonSo, there wasn't much discussion about tearing down LUKS. Makes me wonder if we want to just fork the package with the script we modify and pop in the cosmetic change. Any opinions?20:31
amessersry, i'm not following descussions regularly neither using LUKS anywhere20:33
masonI might transition to using ZFS-native encryption, so LUKS might stop being relevant for me, but as of today the issue in question bites me.20:34
masonOr would without patching.20:34
masongolinux: Did we address the mirror suggestions in the point release? The default entry here is still, but this isn't a brand new installer.20:48
golinuxrrq: parazyd: jaromil;: ^^^  Since there is no us mirror this needs to be changed.  Or does the installer automate that decision making it impossible to exclude a mirror based on location?20:51
golinuxfsr: is this an issue for the -live disks also?20:52
masonI haven't dug through that code to see the heuristics... I don't think the list is autogenerated, so it should be possible to zap it.20:52
golinuxmason:  Thanks for the reminder20:52
masongolinux: It just popped up in front of me as I populate a system for testing LeePen's thing.20:53
masonI'd forgotten otherwise.20:53
golinuxHopefully all involved in building the isos will take note.20:53
golinuxI can't do more than be the messenger20:53
masonLeePen: So, I've been pecking away at the test install, and it's Beowulf now, but I note elogind, libelogind0, and limpam-elogind being held back. I22:21
masonI'm not clueful enough with apt-cache policy to figure out what the conflict is that's keeping them back... Is this expected?22:21
fsmithredmason, I don't remember what package you need to add or remove to make it all better, but aptitude remembers.22:25
masonAh, I'd tried apt and apt-get. I'll try aptitude.22:25
fsmithredsay no, and it will give you other options22:26
masonfsmithred: Good call. It points out libsystemd0 having been installed.22:27
fsmithredthat's what I thought, but I wasn't sure22:28
masonfsmithred: Do we have a recommended way of swapping it out? Or is this why we haven't released Beowulf yet?22:29
golinuxI thought that elogind removed libsystemd022:30
masonfsmithred: Ah, right, aptitude can make the swap. Been ages since I used it.22:31
golinuxI do not have it on my beowulf but not updated for a while22:31
golinuxAs soon as I finish the iso-doc, I'll go do an upgrade22:32
masongolinux: Yeah, it's obviated by elogind, but libsystemd0 seems to get into the upgrade, and apt/apt-get are unwilling to do what they see as breaking the system to swap.22:32
masonaptitude can recommend bigger swaps and make them happen.22:32
* golinux has never user aptitude22:32
masonOnce aptitude has done that, apt and apt-get can be happy again. But this'd be a showstopper for Beowulf.22:33
* golinux goes back to the iso-docs22:33
masonLeePen: Almost there. I'm where I can install build-essentials on Beowulf now.22:33
fsmithredoh, I hadn't checked my recent upgrade. I didn't have elogind or libelogind0. I installed the latter and got both plus libpam-elogind22:41
Centurion_DanI think if we make libsystemd0 a virtual package depending on libelogind0 we can solve that issue22:42
fsmithredThis was a default ascii xfce desktop upgraded to beowulf. I don't know if I was missing any functionality without elogind. Shutdown/reboot worked, but I didn't test removable drives.22:42
fsmithredafk for lunch22:43
LeePenCenturion_Dan: Hi. Are you able to look at the arm64 buildhost? Both rrq and I have been having failure to allocate memory errors.23:07
golinuxmason: I found the answer to your mirror question quite by accident here:
golinux"14) Any of the provided mirrors will be fine. If there is a mirror close to you then you will be redirected automatically when installing packages."23:08
LeePenCenturion_Dan: Is libelogind0 providing libsystemd0 not enough?23:08
golinuxI don't understand how you can be directed to a mirror that doesn't exist!23:08
masongolinux: Right, for reflecting requests. But the issue is that exists as an entry, and that was the issue I thought.23:08
masonLeePen: Not if libsystemd0 gets installed, as happened here.23:09
masonLeePen: aptitude was able to do the "dangerous" swap where the other tools are hesitant.23:09
LeePenmason: great23:09
masonLeePen: Building your packages now, anyway.23:09
masonPlus, done with proper work, so there won't be further distractions.23:10
Centurion_Danmason, would apt install libelogind have solved it?23:10
masonCenturion_Dan: Conflicted with libsystemd0 as I remember. I can do this again on a VM and document the state at each step.23:10
masonCenturion_Dan: Maybe apt takes a DWIM flag. It's always possible I'm missing something. But aptitude was willing to do it.23:11
LeePenCenturion_Dan: I thought that worked too.23:11
masonI'll see about taking a bunch of snapshots of a vm as I push it through the same steps.23:11
masonIt was interesting using aptitude. Been ages. I only grudgingly used it for a while when it looked like dselect was going away.23:14
LeePenBedtime in Europe soon. Back in the morning.23:16
fsmithredoh, mason, I just meant for you to do something like 'aptitude full-upgrade'23:20
masonfsmithred: Oh, I let it resolve the local situation. Find the conflict, suggest an option, !23:20
fsmithredthen it shows you what will be upgraded and gives you other options if there are problems.23:21
fsmithredI never use the curses interface. Looks too much like dselect. :P23:21
masonBut dselect is a good thing! :P23:21
masonI used to like it anyway.23:21
Centurion_Danhmm.  I think for unstable/beowulf we should create the virtual package libsystemd0 that depends on libelogind0.  This would be the easiest and most effective way to solve it.23:22
masonSeems reasonable.23:22
masonCenturion_Dan: Are there other showstoppers? I'd love to have more of a clue about what's actually holding off the release.23:23
Centurion_Dand-i is broken again...23:23
LeePenCenturion_Dan: not quite made it to bed yet! Is libelogind providing libsystemd not enough?23:24
masonLeePen: It does already, doesn't it? Which is why aptitude was willing to swap them, and my system is happy afterwards.23:25
masonHoovering down a desktop environment over DSL... =shudder=23:26
LeePenmason: yes it does. I am wondering if Centurion_Dan is suggesting something else/more?23:26
masonLeePen: Well, the upgrade to Beowulf wanted libsystemd0, so the virtual package would say "sure, here you go" and install libelogind.23:27
masonAs I understand it.23:27
Centurion_DanLeePen: no, because libsystemd0 exists, it get's installed instead of libelogind0 unless it is manually installed.23:27
LeePenOK, I see what you mean. I can look at that tomorrow.23:29
masonLeePen: for this udisks2 testing, do you just want tests against consolekit and elogind?23:30
LeePenYes please. And switching from one to the other.23:30

Generated by 2.17.0 by Marius Gedminas - find it at!